Changes from 2.19 to 2.20

XPLOR Interface

 merged FX-plor functionality. Special thanks to
  Wen Bian, Gerald Stubbs and Don Marvin for making this possible.
  See documentation at
    http://nmr.cit.nih.gov/xplor-nih/doc/current/xplor/
  ------------------------------------------------------------
  added fortlib/specfun/ribesl.f
  ------------------------------------------------------------
   hbdb
      allow zero value of hbprnfrq to disable facility
    ------------------------------------------------------------
      source/genic.f
      source/prototypes.hh
        replaced the DELTI2 subroutine with deleteAtoms, which has
        been modified so that it can be easily callable from the C++
        interface. 
------------------------------------------------------------
  source/xplorFunc.f
    xplor_init: added flag so its not executed twice

------------------------------------------------------------
  arch/Linux_2.2_i686/Makefile.defs
  common/simulation.cc
  source/linuxtime.f
    mods to support gcc/gfortran 4.3 on ia32
------------------------------------------------------------
  source/carb.f
    printcshifts:
      define RMSD as a weighted average of CA and CB shifts
      define NUMBER ans the sum of CA and CB violations
      DECLAR RMSD   --> $RESULT
      DECLAR NUMBER --> $NUMBER
------------------------------------------------------------
  source/etor.f
    added a tolerance parameter dihed_colin_tol:
      the value of cosine of the angle between two bonds above which the bonds 
      are considered colinear. Previously, it was a hard-coded constant.
    it can be set via the command set TOLTorsion <val> end
    the current value can be accessed via the $DIHED_TOL variable.
------------------------------------------------------------
  source/x86_64time.f
  mods to allow compilation using gfortran
------------------------------------------------------------

PASD/Marvin 

  Example Scripts
    eginput/marvin/cvn/refine.py
      Added a final refinement stage.  Refine.py is based on
      eginput/gb1_rdc/refine.py, with filenames changed and only NOE
      and TALOS restraints enabled.   
      
    eginput/marvin/cvn/talos2cdih.tcl
      Also added a short TCL routine to convert TALOS restraints into
      a XPLOR CDIH file, using the rules worked out for Marvin.
      Marvin reads the TALOS files directly, so it doesn't need this,
      but refine.py needs an XPLOR constraints dihedral file.  

    eginput/marvin/cvn/talos/pred/
      Now shipping the individual-residue TALOS output for the CVN
      example, since we're using it in the example structure
      determination. 

    eginput/marvin/cvn/check_talos_predictions.tcl
      Converted CVN eginput scripts to those used for structure
      calculations in paper.   
  
  TCL Interface

    tcl/aeneas_tools.tcl
      -Made writing of .peaks, .shiftAssignments, and .exceptions files
       optional.
      -Added stereopartner recording to standard3d and 4d initMatch
           procs to support stripe routines  
      -Added support in standardJointFilter for applying netfilter
           contact map to different pots than those used to generate it 
      -Bugfixes to correctShiftOffset proc.  In determining output
       report, was comparing calculated offset to minOffsetToApply,
       rather than comparing their absolute values.   
      -Changed flag name in solvent line proc from shiftTolerance to
          just tolerance, to match all the other procs' flags.   
      -correctShiftOffset: minOffsetToApply argument:
          changed default value from 0.02 ppm -> 0.015 ppm
          bugfixes
      -Eliminated calls to initializeLikelihoodsFromFilters in
          standardJointFilter.  These were overwriting the previous
          likelihoods generated from each peak assignment's network
          score.   
      -Added -verboseOutsideSpectrum flags to match*d procs.  Lets
          me control whether complaints about peaks being outside of
          the spectrum are added to the remarks. 
      -Removed symmetryFilter proc
      -Removed primarySequenceDistance proc
      -Removed generateShiftAssignmentPeakPositionLikelihood proc,
          which was never used.   
      -Removed markCrowdedPeaks, which is no longer used
      -Removed stripeCorrection proc, which is replaced by code in
          newstripe.tcl 
      -Removed netfilter proc: functionality moved to netfilter.tcl
      -Removed secStructFilter proc - never used.
      -Updated correctShiftOffset proc, to agree with my old
          codebase.  Now includes flags for minimum number of
          symmetric peaks,  below which no offset will be calculated,
          and minimum offset to apply, below which I judge that it's
          just noise and not worth applying.  Finally,
          correctShiftOffset now throws an error if it does find an
          offset worth correcting, so that the standard*match proc can
          know that it needs to re-match the peaks.   
          This routine might be worth getting rid of entirely, since
          it appears to be relevant for IL-4 only.  My initMatch
          scripts for IL-4 now apply Dan Garrett's hand-determined
          offset corrections to the peak locations before matching, so
          correctShiftOffset isn't even used in that case. 
      -Removed netFilterMini: replaced by code in netfilter.tcl
      -Updated standard3dInitMatch in a few ways:  
           A.  Can now define solvent ranges here, so that solvent
               peaks are removed consistently with any hand-solvent range
               determination 
           B.  Now drops shift assignments whose proton shifts are
               outside of the from and to spectral ranges.  This replicates
               what Dan Garrett does in STAPP, whose parameters don't
               allow folding or aliasing along proton dimensions.   
           C.  Can now define the size of the broad tolerance (in ppm)
           D.  Changed call to correctShiftOffset to handle the
               possibility of an error being thrown (described above). 
           E.  Added -allWeakPeaks flag, which causes all peaks to get
               distance ranges for weak peaks.  Needed because some
               datasets don't have peak intensities available.
           F.  Added -writeEachStage flag, which forces shift assigns,
               peaks, and exceptions to be written at several places during  
               initialMatch, to support statistics gathering on prefilters
           G.  Changed to call new stripe filter code
           H.  Changed standard tight tolerance from 0.03 ppm to 0.02
               ppm, which is what my 2.1.19 code used.   
      -Made corresponding changed to standard4dInitMatch
    tcl/deprecated.tcl
      includes symmetryFilter and primarySequenceDistanceFilter procs- 
      used in old scripts.
    tcl/netfilter.tcl
      -moved the network filter code out into a separate file
      -temporary? change to rollback likelihood
          calc change.  I now just give previous likelihood = 1 for PAs that
          pass, and 0 for PAs that don't.  Was what I had for 1/24/08 marvin
          runs, which seemed to be more accurate. 
      -Changed behavior when raw intrares scores are low.  Was
          previously setting low scores to be equal to the floor
          value.  But that leaves suspicious residues easily able to
          make contacts.  Changed so that low scoring residues get
          their intrares score reset to 1.0, which makes it harder
          (but not impossible) for them to create contacts.  Seems to
          be an improvement, judging by contact maps 
      -Changed netfilter proc to allow it to calculate a contact
          map on the basis of one set of spectra, and to apply it to a
          different set.  This is because 3dN and 4dNC spectra rarely have
          any long-range contacts, and therefore add nothing but noise to
          the contact map calculation. 
      -Bugfix to code that deals with to-from partners.  If a given
          SA has a recorded name for its to-from partner, make sure an
          actual SA with that name exists, because it could have been
          deleted (which leaves the name string describing a non-existent
          SA) 
    tcl/noe_reporting.tcl
      -added completenessSummarizeMarvinResults
         proc, which generates PA likelihoods from the best 10% of
         structures as measured by their local NOE completeness--ie., NOE
         completeness of just the from and to SAs that go into a
         particular PA.  Uses different structures for each PA.  Seems to
         work better in noisy situations, like ThTP pass2. 
      -Fixed overall NOE completeness calculation used in selecting
          converged structures.  Was using mean of each spectrum's
          completeness, which is clearly wrong.  Now I count numbers
          of neighbors and accounted neighbors for each spectrum, sum
          them up, and calcualte overall completeness correctly. 
      -Added new summarizeMarvinResults proc, which ranks structs by #
          peaks nonviolated. Doesn't use completeness.
      -Added nunNonviolatedPeaks proc to support above.
      -Modified chooseBestFraction to support above.
      -Bugfix to old summarizeMarvinResults--was using -inverseBound
          flag to set inverse methyl correction.  Doh, but I don't use
          that proc any more. 
      - Added support for not writing avg PDB, .peaks,
          .shiftAssignments files to newSummarizeMarvinResults and
          completenessSummarizeMarvinResults. 
      - Bugfix to newSummarizeMarvinResults:  Before I create a
          symlink to a converged PDB file with the "converged" tag in its
          name, I now check to make sure it doesn't already exist. 
      - Change to completenessSummarizeMarvinResults.  I now
          activate peakAssigns (for calculating individual SAs' NOE
          completeness) based solely on whether or not they're violated in a
          given structure.  There's no stochastic element to it at all,
          which is appropriate.  I should perhaps eliminate this entire proc
          as a clever idea that didn't pan out, but I'll make that decision
          later. 
    tcl/noe_stats.tcl
      -Changed settings used for updating activation during NOE stats
          calc to prevent SA inactivation, and to explicitly set
          inverse bound 
      -Added new tcl file noe_stats, which contains procs to handle
          activation-sensitive calculation of NOE statistics.   
      -Added procs to calculate 
        number of high likelihood peaks
        number of good high likelihood peaks
        number of high likelihood long range peaks
        number of good high likelihood long range peaks
        number of shift assignments with no high likelihood peak assignments
      -added calls to print those stats as part of the standard newStats report.
    tcl/pdb_support.tcl
      -Changed readPDBremarks proc to require a -fileName flag, for consistency
        with readPDB, which also uses the -fileName flag
      -Added support for writing to the occupancy and b-factor columns
       of PDB files, using names "aux1" and "aux2".
    tcl/import_pipp.tcl
      - Modified to make experiment-specific import routines (eg.,
          read3dCPippPeakTable) optional, and to allow nmrDraw import
          routines to receive and pass on remarks correctly.  All of
          this comes down to running the low-level procs inside of
          eval statements, with the remarks variable names being
          passed along correctly.   
      - Better handling of errors that I catch involving unassigned
          peaks during process3dAssignedPippPeakTable.  I now reset
          errorInfo appropriately, so spurious errors don't appear in
          top-level calls to puts $errorInfo. 
      - I now set the previousLikelihood of peakAssigns in
          process3dAssignedPippPeakTable to 1.0 
      - Added process4dAssignedPippPeakTable, which is entirely
          analagous to process3dAssignedPippPeakTable  
    marvin/ShiftAssignment
      -Eliminated references to EquivPeakAssignmentSet.
      -Bug fix to support removing shift assignments correctly. 
    marvin/MarvinNOEPotential
      - Added functionality to remove all peakAssigns and explicit
          exceptions involving a given shift assignment.  This is
          automatically called when actually removing a shiftAssignment,
          but is exposed to the user as well.  Makes removing shift
          assigns much easier, since it eliminates a lot of TCL-side
          code. 
      - Bug fix involving neighbor list regeneration.  I wasn't
          forcing a neighbor list regeneration when I added or removed
          shiftAssignments or when the inverse caches need updating.   
      - Bug fix in updateInverseCaches.  When I delete a
          shiftAssignment, I clear its toFromPartner's pointer to the
          removed SA.  But I don't clear its toFromPartner SA's
          toFromPartnerName slot.  Now I check whether that name is the
          name of a valid shiftAssign before trying to set the pointer.   
    tcl/shiftassign_tools.tcl
      -Added a proc--dropShiftAssignsInShiftRange, which is used in the
          standardNDinitMatch procs.
      -Bugfix to recordStereoPartnersForShiftAssignments.
      -Added support for recording a shiftAssignment's stereo partner
       in a table.
    marvin/Peak
      Changed chooseAnAssignment method to calc overall likelihood of
      each peakAssignment just as it's done during
      PeakAssignment::updateActivation, and then to normalize them.
      If total likelihood is very low (<0.001), it gives up and makes
      all of that peak's peakAssignments equally likely.   
    tcl/sa_protocols.tcl
      pass2 proc was missing calls to use the original MC scoring
          function and to prevent SAs from being inactivated.  Fixed. 
    tcl/sa_protocols.tcl
      Temporary? change to support changing the values of krama and
          ktalos during pass2 and origPass- 
    tcl/primarySequenceExceptions.tcl
      Code to create inverse exceptions for every SA pair that's
          within a particular number of residues in primary sequence. 
	  Helps ensure that TALOS alone can set the backbone angles, and
          deals with problems caused by few observed NOE contacts  
	  between residues close in primary sequence leading to missed
          contacts.  Primary sequence neighbors often have very  
          limited sidechain-sidechain contacts because of restrictions
          placed by secondary structure. 
    tcl/newstripe.tcl
      New stripe analysis method that avoids using the same peaks as
          targets for multiple shift assignments 
      - Moved code for counting SAs w/ zero targets into its own
          routine, so it can be called either after recording all
          possible targets, or after they've been filtered. 
      - Reorg of prefiltering code (targetIsReasonable).  
      - Now iterates calls to targetIsReasonable during
          filterTargetsForInitialReasonability, to promote convergence.
          Note that this doesn't guarantee identical results each time,
          since it depends on the order in which SAs' targets are
          filtered. 
      - No longer force from SAs to choose targets before the to
          SAs.  Now just choose targets for all SAs, in random order 
      - Changed scoring function from fraction of intrares + bbnseq
          partner SAs that are covered by compatible targets to fraction
          of intrares + bbnseq possible targets that are compatible.   
      - Added optional proc that deletes SAs which have zero targets
    tcl/import_nmrdraw.tcl
    tcl/marvin_lowlevel.tcl
      Change to the flagVal proc, to allow it to match several
          different targets. 
      For example, 
        flagVal $args [list -pot -potList] 
        will match both -pot and -potList calls in an argument list.  
        Needed by several procs in the new stripe filter,
        standardInitMatch, etc. 
    tcl/selection_tools.tcl
      -stereopartnerList proc:  added asn HD21/22 and gln HE21/22 as
       stereopartners.
      -Changed behavior of previousResidue and nextResidue procs to
       always return a single residue string, and throw an error if >1
       next- or previous residue is found.
    tcl/sa_tools.tcl
      -Bugfix to structPrecision.  Was allowing (optional) grabPDBfiles
       call to overwrite main coordinate set before it was copied to
       the comparison set.   
      -Added automatic generation of symbolic links to converged
          structures, so they can be grabbed easily by other scripts
          without having to refer to the avg pdb file's remarks lines.
          Probably useful 
      -Added createViewScript proc, for VMD inspection of converged structures 
      -Changed standard method for generating mean coords to allow vdw
       minimization during cleanCovalentGeom.
    tcl/talos_tools.tcl
      -Folded in individual-residue TALOS input/output from my old codebase.  
       This creates restraints more conservatively, direct from TALOS's
       res_*.tab files for each residue.  The acceptable range is set
       to the observed range of database hits, plus a padding (default
       5 degrees).  There's also a minimum range (default 20 degrees)
       to prevent over-restriction from cases where you  
       have very tightly grouped database hits.  
      
      -This also adds the ability to write out the resulting TALOS
       restraints as XPLOR dihedral restraints.  
      
      -This code also has a bugfix relative to my old codebase.
       Situations where the database hits wrap around +/- 180 degrees
       are handled gracefully.  This was causing surprisingly large
       numbers of bad (or extremely broad) TALOS restraints  
       in my old codebase.  
    ------------------------------------------------------------

------------------------------------------------------------
Python
    changed default Python version for output from 2.4 to 2.5
    ------------------------------------------------------------
    new:  python/symmetry.py
        contains test for dimer C2 symmetry
    ------------------------------------------------------------
    added distance symmetry potential: 
      same as the XPLOR NOE distance symmetry restraints for normal
      Simulations, but enforces symmetry between EnsembleSimulation
      members.
    ------------------------------------------------------------
    atomSel module
      added asUnion/union function
    nmrPot/posSymmPot.cc
       -fixed bug in gradient
       -now throw an exception if there are fewer than 3 equivalent atom sets
       -changes to PosSymmPot mostly to expose internals so that the 
        translate/rotate operations could be repeated externally
        rationalized return type of equivAtomSelPairs and equivAtomPairs
       -added rotMat, centroid and translatedCoords accessors
    nonBonTools.py
       vdwViolations now takes extra arguments to optionally determine
       violations involving only selected atoms.
    python/xplorPotTools.py
      PotData.__init__:
        added try...except clause for getting violation info
          motivated by distance symmetry NOE restraints for which the noe
          print statement doesn't work.
      now print out XplorPot::scale in analysis (.viols file)
      now get, print rmsd, violation info from the CARB term
    ------------------------------------------------------------

    rdcPot
      added gyroA/B output to info()
  
    atomSelAction:
      RMSD: constructor can also take an AtomSel argument
      Fit.translation now returns a Vec3, not a tuple.
    ------------------------------------------------------------
    python/simulationTools.py
      StructureLoop:
        -rmsd to average structure is now calculated
         correctly. Previously, rmsd was calculated to unregularized structure.
        -constructor: added averageRegularize flag to allow disabling
                     regularization. 
        -added the structureNum method
        -averageFilename now supports the SCRIPT replacement of pdbTemplate.
        -fixups for EnsembleSimulation calcs:
           -now call barrier in constructor to clean up list of
            running processes early 
           -run(): made cpuTime calc EnsembleSimulation-safe
      StructureLoop.genAveStats: allow for missing files when
         generating statistics (instead of crashing) 
      StructureLoop.run() now sets a number of members to the instance
         (self) object, including cpu time, calculated structure precision and
         restraint statistics. Also, it now returns the instance object.
      StructureLoop.writeStructure:
        set the default potList and altPotList arguments from
        StructureLoop's averagePotList and averageCrossTerms, respectively
      calcAverageStruct:
        now takes new arguments potList and regularizeSteps, used for optional
          regularization by minimizeRefine
      minimizeRefine:
          fixed typo in clause which ensures covalent energy terms are present.
      testGradient:
          -added a text header for EnsembleSimulation gradient tests
          -fixes for when dealing with EnsembleSimulations: it should
           now actually work correctly and now supports xyz components
      convertToPotList: can now take a bare potential object
    ------------------------------------------------------------
    python/solnScatPotTools.py
    python/solnXRayPotTools.py
      added Xiaobing Zuo's nucleic acid volume values- can optionally 
        activate these values.
      added Mn,Zn to supported metal atoms
      create_solnXRayPot: changed the default selection from not hydro
    ------------------------------------------------------------
    ivm
      -added untested rotate2 node type
      -fixes s.t. IVM.hinge() actually works with previous argument types
      -optimization of breakAllBondsIn method- moved loops to a
        C++ extension defined in the .i file.
      -added simulation() accessor
      step():
        now check if newTimeStep is smaller than minStepSize and throw an
        exception if so.
      -additional optimizations
    ------------------------------------------------------------
    python/residueAffPotTools.py
      added 1999 Mizagawa/Jernigan contact potential 
      create_ResidueAffPot: 
        added optional argument interdomainContacts.
        fix justification (to left) of segname
        now has optional potentialName argument to switch
          between different potentials
    ------------------------------------------------------------
    python/selectTools.py
      numResidues:
        fixed default argument (so a null selection correctly selects
        zero atoms) 
      added getSegids() helper
      added convertToAtomSel() helper
    ------------------------------------------------------------
    avePot:
      fixed gradient error which sometimes would double the gradient
    ------------------------------------------------------------
    python/psfGen.py
      seqToPSF: 
        right justify space-padded segid
        the length of segName is now handled correctly.
      deduceSeqType:
        will now throw an exception if a residue name is longer than
        four characters 
      pdbToSeq:
        added helper code to catch the case of truncated ATOM records:
        short lines such that the segid has fewer than 4 characters.
    ------------------------------------------------------------
      python/regularize.py
        fixupCovalentGeom:
          -now catch and ignore small timestep errors from the IVM
          -new algorithm which works residue-by-residue
           and uses fast SetProperty and new AtomSel helper functions.
          -vdw fixup
          -added fixupLoop_xplor: now do residue-by-residue fixup in XPLOR
        addUnknownAtoms:
          now just exit if there are no unknown atoms
        fixupCovalentGeom: tighten XPLOR TOLT parameter for this routine
        simple fixes to covalentMinimize
    ------------------------------------------------------------
      python/protocol.py
        added initHBDB helper function
        matchInexactEntry:
          added new name transposition rule.
          matchInexactAtomEntry:
        import regularize.CovalentViolation
        initCoords matchInexactAtomEntry: 
          add strictResNames flag argument
        moved covalentMinimize and fixupCovalentGeom to regularize.py
          genExtendedStructure: now use fast atomSelAction.SetProperty
      genExtendedStructure:
        changed the default atom selection to exclude RDC axis atoms. This
        so it will now do the right thing when called after the RDC terms
        are set up.
      pdbLocation:
         switched to new .gz ftp location
      downloadPDB:
         now use urllib2
      loadPDB:
        fixup resetting of current simulation
      added addUnknownAtoms_fast, and call to it from addUnknownAtoms if the
        missing atoms are separated by a single bond from a known atom
    ------------------------------------------------------------
      pdbTool
        added strictResNames flag and ability to require strict residue naming
        changed exception type returned by PDBTool::read to IOError
    ------------------------------------------------------------
      atomSel
          added helper functions intersection(), notSelection() and union()
          which create pseudo selections
    ------------------------------------------------------------
      common/xplorWrap.cc
        deleteAtoms:
          added call to XplorSimulation::syncFrom after deletion
    ------------------------------------------------------------
      simulation
        all Simulation objects are now reference counted
        Simulation::deleteAtoms:
          the argument is now a list of atom indices instead of a
          selection string.
          the Python .i files added an indirection to keep the
          previous behavior. 
        added atomMassArr/setAtomMassArr accessors
    ------------------------------------------------------------
      common/atomSelAction.cc
      python/atomAction.py
        moved SetProperty from atomAction to atomSelAction for optimization. 
        Renamed previous definition to SetPropertyArr
    ------------------------------------------------------------
      common/xplorWrapProc.cc
        freeAtomData: bug fix:
          now use dynamic_cast and check result when down-casting to obtain a 
          XplorWrapProc.
    ------------------------------------------------------------
      python/socketComm.py
        collect: added diagnostics in case of readData failure
    ------------------------------------------------------------
      added posDiffPotTools
       a potential term to help evaluate structural RMSD
    ------------------------------------------------------------
      selNBPot
        added CENTER aveType to selNBPot
    ------------------------------------------------------------
      restraintStats
       TermEntry now has methods to calculate mean, dev of energy,
       rmsd and viols 
    ------------------------------------------------------------
      added python/maxLikelyFit.py
       module for molecular fitting based on maximum likelihood estimate.
       written by Robin Thottungal
    ------------------------------------------------------------
      jCoupPot
        better diagnostics if bad selections are encountered

------------------------------------------------------------

Low Level Libraries
    ------------------------------------------------------------
      CDSlib/cdsList.hh
        sort method: try to make a default comparer argument
    ------------------------------------------------------------
      CDSlib/rc_ptr.hh
        allow proper cleanup of pointee.
        no longer set counter member to zero.
        copy constructor: made safe for null counter members
        added methods for manual reference count manipulation:
          incr, decr count
        added methods for casting up/down TYPE class heirarchy:
           cast, dyn_cast
        mods so that destructor zeros counter member

Helper Scripts
------------------------------------------------------------
  calcTensor, calcETensor, calcDaRh have been moved into the bin directory
  added mleFit and domainDecompose helper programs
  seq2psf:
    added the -cisPeptide option and usage test
  targetRMSD:
    add check that all atoms used for fitting are known
  calcTensor:
    fix crossTabs initialization type
    added -crossValidate option

  bin/xplor
    -handle the case in which the machines list starts with localhost, but
     also contains nonlocal machines.
    -formalize and document arch/equivList format
  tcl/findTCL
    support for recent versions where TCL_INC_SEPC is replaced by
        TCL_INCLUDE_SPEC 

Example Scripts

  -added validation scripts
     these check that the calculated results are of suffient quality.
     These can take a long time to run. Run with 
       eginput/validate
 
   -eginput/runAll
       added -keep command-line option

    ------------------------------------------------------------
   eginput/gb1_rdc/
    * added analyze_anneal.py, analyze_refine.py for analysis, validation
    *refine.py
      -added SCRIPT to averageFilename
      -accept function:
         loosened tolerance on rdc rmsd
      -properly ramp the Vgyr scale factor
      - added a cross-validated PosDiffPot term to compute rmsd to the crystal
        structure.
      - bumped the number of calculated/analyzed structures to 20, and added a
        -quick flag to reduce it back to 3

    *anneal.py
      -now start from randomized torsion angles instead of extended structure
      - anneal:
       - initialize coords from model.pdb - these coords are not used in calcs
       - remove initial calcTensorOrientation- but add it after extended 
         structure is generated
       - set averageContext in StructureLoop
    *analyze.py
      - allow for missing structure files (instead of crashing)
   ------------------------------------------------------------
     eginput/dna_refi/ensemble.py
        fixup selections: hydro --> name H*
   ------------------------------------------------------------
      eginput/PSF_generation
      *testNAG.py
        increased maxFixupIters
      * added addAtoms2.py
      * added writeChainID.py example


		     Changes from 2.18 to 2.19

new helper script:
  bin/targetRMSD - to compute rmsd to a known structure

arch/equivList - Mac platform
      updated to run Darwin_9, confirmed by user feedback

added parallel.txt
      write-up describing parallel structure calculation capabilities

configure script
      improve error message for case appropriate xplor binary is not present.

XPLOR interface:
      removed the BUFMAX parameter from comand.fcm
      a consequence is that the rotating command buffer no longer rotates:
        it just gets larger
      added two array helpers:
        assign_char_2d and get_char_2d
      removed the NDSMAX parameter - associated arrays are now on the heap
      removed the MXRTP parameter
      removed the MXRTI parameter
      removed the NHBMAX constant parameter
      fixed ASSIgn docs for xdipo/dipo potential
      added xadc XPLOR term contributed by G. Bermejo (reference ???)
 
  source/parser.f
      changes to sizes of heap-allocated strings. Probably not needed.

   XPLOR angle, bond, dihedral and improper violations:
      added NUMBER declar'd variable

Python Interface

  gyrPotTools:
       updated Vgyr ranges to match those in the paper
        added Rg targetType - needs gradient, testing of Rg targetType
  
  pdbTool:
      added ability to write to the ChainID field (by setting the 
      writeChainID flag).
      PDBTool::read():
        better support for a subset selection- these atoms are no
        longer added to the unplaced list. 
  
  protocol:
      loadPDB: added simulation argument
      addDisulfideBond: fixed to make working, plus docstring
      initDynamics:
        fix so that eTolerance isn't specified as negative if bathTemp argument
        is omitted.
      initCoords: now allows selection to be a plain string
      fixupCovalentGeom:
        changed default angle and dihedral tolerances from 5 to 2
        degrees. This picked up some problems in less-used parameters.
      matchInexactAtomEntry:
        handle case when occupancy or Bfactor are invalid numbers

  xplorPotTools:
      now get number of XPLOR restraint terms from XPLOR NUMBER variable
      added restraints class support for the ANGL XplorPot term

  python/psfGen.py
      moved addDisulfideBond from protocol to psfGen w/ compatibility call

  spline:
      added support for periodic splines
      added Spline2D - bicubic interpolation, w/ gradient

  vec3:
      __neg__: fixed so that it doesn't have side effects

  dihedral:
      overhaul of Dihedral class
      changed return units from degrees to radians
      added derivs method
      changed constructor (old constructor is emulated by Atom typemap)

  densPotTools:
      create_DensPot:
        fixed location of an import statement

  added cstMag potential for refinement against chemical sheilding 
     tensor magnitudes. Reference forthcoming.

  nonBondTools:
      allow atom names to contain the +'' character

  simulation:
      added Simulation::deleteAtoms

  ivm:
      ivm.hinge: change arguments to be more sane.
                 backward compatibility maintained.


  added support for Python2.5

Example scripts

  eginput/PSF_generation/testDiS.py
      added disulfide test/example

  eginput/gb1_rdc/bicelles_new_hnc.tbl
      removed one rdc due to consistent violations
  eginput/gb1_rdc/refine.py
      accept: loosened required rdc tolerance

  updated eginput examples in eginput/gb1_rdc/


  eginput/protG/calcDaRh.py
      small fixes 
        CH normtype should now work
        modePowder should now be correct
        noise argument should be better handled
      documentation fixes

  eginput/protG/calcTensor.py
        - added normType option for RDC scaling
        - fixes to correctly handle multiple rdc experiments
        - added rdcWeights option to correctly weight multiple experiments
        - added the following command-line flags:
           -psf, -useDistance, -maxDa
        - fixed bug in which tensor atoms were improperly reset before
          calculating rmsd.

  eginput/PSF_generation/addAtoms.py
      added call to fixupCovalentGeom

Parameter/Topology Changes

  toppar/ion.par and toppar/ion.top:
      changes to avoid conflicts with protein.top/par

  toppar/nucleic-1.1.par
      nucleic base proton angle parameters:
        adjust angles proton+heavy atom angles so that they add up to
	360 degrees. Previously, there were always violations of >2 degrees.


Internal Variable module:
    fixed handling of diatomic molecules

    resetCM():
          now perform reset if fixed atoms are present - previously it just
        exited if there were any fixed atoms.

    corrected center of mass calculation:
      this improves algorithm stability for the case of nonuniform masses.


Marvin/PASD 

  tcl/noe_selecting.tcl
      Changed definitions in isShortRange and isLongRange to do
      exactly what the C++ code in PeakAssignment.cc does, in order to
      eliminate confusion. 

  eginput/marvin/cvn/convert_restraints_to_xplor.tcl
      removeLowLikelihoodPeakAssignments:
        fixed name of the -cutoff flag

LICENSE
      added government rights clause from A. Hubbs

switched from cvs to bzr version control software.

documentation fixes, additions


		     Changes from 2.17 to 2.18

Helper scripts

  -bin/seq2psf
        added the -protonateHIS flag
  -added bin/targetRMSD


C++ infrastructure

  CDSlib/cdsComplex.hh 
        fixed up pow(complex,pow)
        added functions operator*(int,complex) and pow(complex,exponent)

  CDSlib/cdsMatrix.cc 
        now support multiplication with GenMatrix; SymMatrix*vector now works
        added rawCopy method

  CDSlib/cdsVector.hh 
        splitRep: made this method public

  CDSlib/fft.cc 
        disambiguate ::pow call for Irix
        fixed call to ::pow
        added wrapper class for FFTPACK

  CDSlib/fft.hh 
        added wrapper class for FFTPACK

  CDSlib/fftpack.hh 
        removed redundant consts
        int --> long conversion: needed for 64 bit arches
        added wrapper class for FFTPACK

  CDSlib/matrixTools.cc 
        determinant:
          reworked to have saveData- for optimization
          explicitly coded size 2 and 3 expressions - much faster than lapack calls.
        added function determinant (from PLU decomposition) and tests
  CDSlib/symMatrix.cc 
        now support multiplication with GenMatrix; SymMatrix*vector now works
        added missing #include
  CDSlib/vec3.hh 
        crossMat: reworked to use only constructor

  intVar/publicIVM.hh 
        added recenterLargeDispl to PublicIVM: used in
        syncPos. Disabled by default.

  common/atom.cc 
        atom:
          pos, mass, vel, fric read accessors have been inlined.
          This required #including simulation
        simulation:
          which required un-inlining Simulation::atomByID and rearrangement of
          #includes
        made usual constructors inline

  common/atomSel.cc 
        AtomSel: moved binary methods out to stand-alone functions
          intersects --> intersect
          isMonomericallyEqualTo --> monomericallyEqual
          primarySequenceDistanceTo --> minResidDiff
        Added primarySequenceDistanceTo <anotherAtomSel> function.  This is broadly
        useful, so it belongs here.  Need to add a test for it.

  common/atomSelAction.cc 
        updated Fit and RMSD to use AtomSels for fitting, comparison

  common/mmapAlloc.cc 
        added the deleteFile function
          so that the mmapped file can be deleted regardless of order called.

  common/pdbTool.cc 
        better treatment of altloc and icode entries:
          nonblank icodes are now ignored
          altloc enties in allowAltLoc are read, otherwise entry is ignored.
        added commented-out check of residue name
        write: check if atom coordinates are within range (valid). Warn if not.

  common/pot.hh:
        calcEnergy() now returns float_type instead of EnergyReport

  common/simulationWorld.cc 
        added memoryUsed function

  common/xplorSimulation.hh 
        added clone argument to constructor - if True, the new
        XplorSimulation will be a copy of the current one. If False,
        if will be a brand new XplorSimulation, with no atoms, etc.

  common/xplorWrap.cc 
        fixed bug: SELRPN messages were printed when XplorWrap->select called
        added the qatmchk argument to the XPLOR energy() subroutine.
        This defaults to .TRUE. for all XPLOR use. Within the C++ interface it is
        .FALSE.
        This is for performance in addition to provide proper (not crash)
        behavior, particularly when gradient minimizing.

  common/xplorWrapProc.cc 
        XplorWrapProc::freeAtomData:
          make sure pointers are correctly mmap or free - deallocated
        exec'ed children are now started -quiet -- banner is not printed
        disable master_exit_handler:
          it kills the slaves, and they don't get a chance to clean up.
        two bugs fixed:
          the master process now has a xplorDerivs instance variable - it had
          been using the static variable, leading to problem when multiple
          XplorWrapProcs were used.
        shutdown/cleanup change:
          now only the slave deletes sharedData. It seems to be cleaned up
          anyway when the slave exits- causing semaphore errors when the
          master had tried to clean up.
        added the clone argument to the constructor:
          if false (old behavior) a new xplor process is exec'd, if true
          the existing forked process info is used.
        static variable firstCall in function allocSharedData replaced by
          a variable in the instance-specific sharedData structure -
          this so that calls to XplorSimulation after the first one would be
          successfull.
        xplorwrapproc_slave:
          added a wait loop so that the slave process waits for the master at
          startup
        fixed bug: SELRPN messages were printed when XplorWrap->select called
        added the qatmchk argument to the XPLOR energy() subroutine.
        This defaults to .TRUE. for all XPLOR use. Within the C++
        interface it is .FALSE.
        This is for performance in addition to provide proper (not crash)
        behavior, particularly when gradient minimizing.

  common/xplorWrapProc.hh 
        two bugs fixed:
          the master process now has a xplorDerivs instance variable - it had
          been using the static variable, leading to problem when multiple
          XplorWrapProcs were used.
        shutdown/cleanup change:
          now only the slave deletes sharedData. It seems to be cleaned up
          anyway when the slave exits- causing semaphore errors when the
          master had tried to clean up.
        see the commit message for xplroWrapProc.cc

Example scripts:

  eginput/PSF_generation/testPDB2PSF 
        mods for disuse of seqres, accompanying psf fixes

  eginput/gb1_rdc/README 
        added targetRMSD.py script, test

  eginput/gb1_rdc/analyze.py 
        updated to split NOE restraints and turn off ANI VdW interactions
        now report a proper VdW energy

  eginput/gb1_rdc/anneal.py 
        updates:
          note proper # structs
          now generate new extended structure for each structure calculated
        added missing (optional) cart_cool definition
        fixed typo

  eginput/marvin/cvn/cvn_n15noe.shifts 
        Corrected some typos in 15N shift table:
        1.  Several residues had two lines of, eg., HA1|HA2 instead of
            HA1|HA2 and HA2|HA1.  Fixed. 
               Residues affected:  2 5 6 15 27 28 45 65 78 96
        2.  Also changed selections for phe/tyr deltas & epsilons from
            CD1|CD2 to CD*, since they're almost certainly averaged.
               Residues affected: 4 9 29 54 80 100
        3.  Finally, note that previous version had a typo in
            definition of Y29 epsilon carbon: 
            Was CE1 | CD2.  Changed to CE#, which fixed the problem.
        
        Note that this shift table should be the input for TALOS, not
        the 13C one (which lacks entries for backbone carbonyls, for
        some reason). 

  eginput/marvin/cvn/initMatch3dC.tcl 
  eginput/marvin/cvn/initMatch3dN.tcl 
        Updated to reflect name change of correctShiftReferencing proc to
        correctShiftOffset

  eginput/protG/README 
        added calcDaRh.py - determine Da, Rh in absence of structure info

  eginput/protG/anneal.inp 
        rework parallel startup logic again - for case of num_procs>numStructs
        fixes to avoid non-integer structure count

  eginput/protG/calcDaRh.py 
        added calcDaRh.py - determine Da, Rh in absence of structure info

  eginput/protG/calcTensor.py 
        will now (optionally) print out back-calculated RDCs
        fitSel can also now be specified on the command-line
        updated to support command-line options


  nmrPot/Makefile 
        added gyration tensor volume potential term
        added selNBPot/ residueAffPot--
          empirical potential to capture hydrophobic forces.
          Current potential based on a K. Dill contact map.
        added planeDistPot
        approximate molecular volume calculation from analytical surface area

  nmrPot/gyrPot.cc 
        added square well potential
        gyration tensor volume potential term

  nmrPot/gyrPot.hh 
        added square well potential
        gyration tensor volume potential term

  nmrPot/neighborBin.hh 
    deal with case where too many bins would be created - atoms are very far 
    apart 
    changed template parameter from pointer-to-atom to atom instance.
    also added an ATOMLIST template parameter.
    no longer refer to an atom radius member- instead just pass a simple
    spacing argument.

  nmrPot/noePot.cc 
    added support for ASSIgn ... OR restraint syntax
    initial changes to support the TO keyword in the ASSIgn statement:
      in the restraint class, replaced sel1, sel2 members with selPairs()
      accessor
    
      The python layer includes a compatibility layer such that sel1 and sel2
      continue to function.
    now throw an error if the ASSIGN .. OR syntax is detected

  added planeDistPot distance-to-plane restraint potential term

  nmrPot/selNBPot.hh 
        added dSwitch
        bug fixes in neighbor determination
        added selNBPot/ residueAffPot--
          empirical potential to capture hydrophobic forces.
          Current potential based on a K. Dill contact map.

  nmrPot/surfaceArea.cc 
    added null destructors
    optimizations
      changed many instances of rc_Neighbor to SA_Neighbor*
       removing this indirection lead to a large speed-up
     did same with rc_Vertex:
       no big gain from that.
    comment-out updateNeighbors: not used for moveTol==0
    re-added neighborBin for optimization
    overlap: added moveTol argument
    tried to make nonzero moveTol work. No luck yet.
    replaced code with transpose with left-vector multiply
    now derivs from ModifiedBase
    removed TouchingList::init
    
    various optimizations, using const where possible,
      using raw pointers instead of ref counted pointers
    
    bug fix: now initialize sim member
    
    cache neighbor list in new neighbor member
    
    singleAtomDerivs:
      add optional scale parameter
      avoid divide-by-zero error by adding small value to Omega
    
    getOverlappingAtoms:
      use touchingList if cRadius allows it
    
    added A member- previously determined surface area. Used to avoid 
    unnecessary (possibly dangerous) derivative calculation.
    
    optimization (?) to outerProd
    Vertex constructor: added initializers
    now expose more info: Omega angle

  nmrPot/volume.cc 
        added null destructors
        cache atom.pos()
        coded derivatives of tetrahedra in nonlazy fashion
        changed deriv support from VecVec3 to DerivList
        added sa_saveData member
        added derivative info
        approximate molecular volume calculation from analytical surface area

  nmrPot/volume.hh 
        added null destructors
        changed deriv support from VecVec3 to DerivList
        added sa_saveData member
        added derivative info
        approximate molecular volume calculation from analytical surface area

Python Interface

  python/atomAction.py 
        SetProperty:
          can now specify array of values.
        added getProperty function.
        added function: centerOfMass
        randomizeDomainPos: fixed error in CoM calc

  python/cdsMatrix.i 
        support inter-symMatrix/matrix multiplication
        added much SymMatrix support -
          multiplication by matrix, vector, svd, inverse
        added space before trailing template > for later swig versions
        SymMatrix:
          fixed constructor, so that size-only can be specified.
          removed get method
          changed set method to use underlying method.
        moved cdsComplex stuff to cdsComplex.i

  python/cdsTypeMaps.i 
    removed unused variable from DEFAULT_REF_TYPEMAP
    DEFAULT_REF_TYPEMAP - Python-to-C++:
      add attempt to directly copy any underlying C++ object to the target
      C++ object without going through the fromPy machinery - this is an
      optimization.

  python/cdsVector.i 
        now references the correct documentation
        added CDSVector<DComplex>

  python/cdsVector_template.i 
        added space before trailing template > for later swig versions
        added __str__, __mul__, __div__

  python/csaPot.i 
        allow positive values of Da, and changed sign convention
        added inline functions for computing CSA powder patterns

  python/csaPotTools.py 

  python/densPotTools.py 
        fixup to correctly count the mass
        added explicit destructor, call to base class destructor
        create_SurfaceArea:
          added radiusType argument, support for heavyatom radius parameters
        
        support for XplorWrapProc XplorSimulations
        create_DensPot: don't crash if number of attached protons is unexpected
        create_DensPot:
        
          now disable convex hull by default
          explicitly set moveTol to zero (will crash with any other value)
        replaced calcEnergyAndDerivs with calcEnergyAndDerivList
        changed default atom selection
        corrected capitalization of createDensPot
        corrected default target value
        fixup for atom radius setup.
        make timing info optional
        density potential and test

  python/derivList.i 
        added __get/setitem__ methods
        get method return val is no longer const - impt. for Python pot terms

  python/fft.i 
        added fft module

  python/findPython 

  python/findSwig 

  python/gyrPot.i 
        gyration tensor volume potential term

  python/gyrPotTools.py 
        create_GyrPot:
          added softScale, hardScale arguments
        updated documentation
        now create two terms - hard and soft.
        change default lower hard range from 14.5 to 14
        replaced tag atom selection with selectTools.numResidues
        added square well potential
        create_GyrPot:
          added resScaleFactor argument, changed default value
          reduced default scale factor to 1
        analyze:
          output format fixups
        gyration tensor volume potential term

  python/hybridMC.py 
        use getXplorSimulation instead of xplor.simulation

  python/ivm.py 
        cleanup
        run:
          now checks for atoms with large displacements before dynamics.

  python/mat3.i 
        added inverse

  python/minimize.py 
        conmin:
          added stepsize argument
  python/noePot.i 
        initial changes to support the TO keyword in the ASSIgn statement:
          in the restraint class, replaced sel1, sel2 members with selPairs()
          accessor
        
          The python layer includes a compatibility layer such that
          sel1 and sel2 continue to function.

  python/noePotTools.py 
        create_NOEPot:
          added splitRestraints, deltaResidLR arguments, functionality

  python/noeTest.py 

  python/noeTest.py.out 

  python/nonBondTools.py 
        update to work with XplorWrapProc
        now allow chemical types to contain underscores
        readXplorRadii:
          updated to correctly use 1-4 list
          now read nonbonded exclusion list from the PSF and use that too
        vdwViolations:
          change default value of radiusScale from 1 to defaultRadiusScale=0.8
  python/planeDistPot.i 
        distance-to-plane restraint

  python/planeDistTools.py 
        update to work with XplorWrapProc
        helper functions for distance-to-plane restraint

  python/protocol.py 
    fixes to xplor manual URLs
    added helper function:
      initCarb13
    loadPDB:
      added optional string argument
    matchInexactAtomEntry:
      added check for residue name mismatch
    loadPBD: added verbose argument
    added pseudoResNames:
      residue names to ignore when counting unreadable pdb ATOM entries
    initCoords:
      maxUnreadEntries is now a tuple, and value increased.
    downloadPDB:
      now remove temporary .pdb file
    initCoords:
      overload maxUnreadEntries to allow number to be specified as
      fraction of the number of residues.
    readPDB renamed to downloadPDB
    loadPDB: now has two arguments, to load from file or url
    matchInexactAtomEntry:
      now bomb if a nonnumeric resid is encountered
      altloc- now treat with PDBTool.allowedAltLoc
    initCoords:
      added maxUnreadEntries argument - to bomb on psf/pdb mismatches
    added loadPDB, readPDB helpers, to aid in reading PDB entries from the net
    reverted to previous version of genExtendedStructure
    genExtendedStructure:
      added genWithoutProtons argument to trigger an alternate algorithm
      in which a sub-XplorSimulation was created without protons and first
      fixed-up. Then protons were added with hbuild. The idea was that
      this would be a faster algorithm. Unfortunately, it turned out to
      not be the case.
    fixupCovalentGeom:
      now check that the same pot combination is never used twice in a row.
      added an additional verbose level.
    genExtendedStructure:
      added maxFixupIters argument

  python/psfGen.py 
    pdbToSeq: new arguments:
      useSeqres: defaults to false, better to read sequence from ATOM
                 records
      useChainID: useChainID - overrides segment name by default
    also: now terminate sequence determination when an ENDMDL record is found
    seq2PSF: added support for n-terminal ACE residue
    fixes so that a nondefault XplorSimulation can be used
    residueTypes changed from tuples to lists

  python/residueAffPotTools.py 
        included simulation argument in AtomSel constructor
        create_ResidueAffPot:
          added sequentialCutoff argument
        analyze:
          fixed attractive reporting.
          added repulsive report
        SelNBPot/ ResidueAffPot documentation
        added selNBPot/ residueAffPot--
          empirical potential to capture hydrophobic forces.
          Current potential based on a K. Dill contact map.

  python/restraintStats.py 
        summarizeTerms:
          now break out terms in PotLists in a separate listing, s.t. the
          total energy is approximately the sum of all terms.
        added number of restraints column to the summary output.

  python/selectTools.py 
        numResidues:
          removed spurious print statement
        updates to rigidNucleicSelections
        added convertToAtom

  python/simulationTools.py 
        testGradient:
          added tolerance and epsilon arguments
          gradient difference is now scaled by gradient magnitude
        testGradient:
          added optional components argument
          now output is better formatted, and has a header.
        StructureLoop constructor:
          logic to better handle case where # procs > # structures. Avoid
          error message and print informational message.
        added comment concerning refactoring StructureLoop.genAveStruct
        testGradient: updated the docstring

  python/surfaceTools.py 
        create_SurfaceArea:
          added radiusType argument, support for heavyatom radius parameters
        
        support for XplorWrapProc XplorSimulations
        quiet XPLOR output
  python/varTensorTools.py 
        update to work with XplorWrapProc
        calcTensor_ensemble: added svdTolerange parameter; _analyze:
        fix problem with axes exactly alon xyz direction 
  python/xplorSimulation.i 
        getXplorSimulation: added default argument

Fortran source

  added the DIRS XPLOR term from N. Tjandra

  RAMA potential term:
    in files:
     databases/torsions_gaussians
     databases/torsions_quarts/
     databases/torsions_raw
           Changed delphic torsion setup scripts.  In code that selects
           free and disulfide bonded cysteines, I no longer use the
           existence of an HG atom as an indicator of whether or not a
           residue is in a disulfide.  Rather, I actually look for an SG
           atom that's bonded to the current residue's SG.  Thus,
           situations where the HG is missing, but no actual disulfide
           bond exists, will not be treated as if they were disulfides,
           eliminating a small source of errors.
           
           In cases where users have deleted the HGs and implemented the
           disulfide using NOEs, the DELPHIC torsion restraint will treat
           the cysteines as if they were not disulfide bonded.  Since
           NOE-based disulfides are not likely to have correct CB-SG-SG
           bond angles, there's no point in trying to have correct
           CB-SG-SG-CB torsion angles.
           
           This change to the setup scripts also eliminated a source of
           These corrections may result in minor changes to xplor output
           of various scripts!  Specifically, they will probably
           eliminate a few lines of "no atoms selected" style errors in
           the output of the DELPHIC torsion setup calls.
           
           MANY THANKS to Chun Tang from GMC's lab for catching this!
   

  source/darwintime.f 
        split out GETSYS into separate file for x86, ppc

  source/energy.f 
    added the qatmchk argument to the XPLOR energy() subroutine.
    This defaults to .TRUE. for all XPLOR use. Within the C++ interface it is
    .FALSE.
    This is for performance in addition to provide proper (not crash)
    behavior, particularly when gradient minimizing.

  source/nbonds.f 
        upinb: suppress normal printing of 'Ran out of space' message

  source/update.f 
        suppress SCRATC-warning: message if WRNLEV==0

  source/noe.f 
        Eliminated weird status message.  Whenever R-6 averaging was
        selected, xplor would print "R-6 averaging.", for no apparent
        reason.  Wouldn't do that for any other averaging type.

        Updated documentation of the help message for the
        NOE ASSIgn statement to reflect the existence of the 
        ASSIgn ... OR syntax. 
        

Marvin/PASD updates

  marvin/MarvinNOEPotential.cc 
        Added hasExplicitInverseException routine, and added == operator to
        ExplicitInverseException

  marvin/Peak.hh 
        Added isCrowded flag to Peak class

  marvin/PeakAssignment.cc 
        Added isBackbone flag, isBackboneSequential function.
        Moved primarySequenceDistance routine to ShiftAssignment class
        to make it available in more cases.

  marvin/ShiftAssignment.cc 
        Moved primarySequenceDistance calcualtion from PeakAssignment to here

  tcl/aeneas_tools.tcl 
    change set A
      1.  Bug fix to reportUnassignedAtoms:  No longer complains about
          missing proline Ns 
      2.  Added -bbn-seq flag to PSD filter
      3.  Added markCrowdedPeaks proc, for use by stripeCorrection
      4.  Updated stripeCorrection for markCrowdedPeaks
      5.  Updated stripeCorrection to handle cases where no target is
          expected (eg., in 3dN with only intraresidue peaks used as targets)
      6.  added netFilterMini, the poorly-named residue-pair version of
          the network filter 
      7.  Added convenience routines for processing:
              standard3dInitMatch
              standard4dInitMatch
              standardJointFilter
      8.  Added passSummaryJointFilter, which I'll probably remove

    change set B
      1.  Made match*d tolerant of shiftAssigns that don't have
          chemical shift values. 
      2.  Gave match*d ability to not use shiftAssigns with low
          previousLikelihoods. 
      3.  Improved reporting of peaks outside of the spectrum in match*d.
      4.  Gave recordUnfoldedPositions the ability to work with a
          subset of a pot's shiftAssignments.
      5.  Corrected a not-yet-tickled bug in removeDiagonalPeaks.  If
          you're enforcing heavyatom positions being near the diagonal
          as well (only occurs in 4dCC spectra), then I forgot to
          multiply by 1/sqrt(2). 
      6.  Explicitly set previousLikelihood of newly created
          ShiftAssigns to 1.0. This doesn't actually change behavior,
          since the C++ default was 1.0.
      7.  Changed networkFilter to not count pairs of SAs that are
          to-from partners by default, since they don't generally have
          peaks or peakAssignments corresponding to them. 

  tcl/atomSel.i 
    AtomSel tcl wrapper code:
      extend TCL wrapped class with intersects method to minimize code change. 

  tcl/import_nmrstar.tcl 
    Added two features:
      1.  Can handle cases where shift entries have an author_residue_id code.
      2.  Expanded automatic mapping of nmr-star atom names to xplor
          atom names. There should be very little need for editing
          nmr-star shift tables by hand any more.
  tcl/marvin_lowlevel.tcl 
        reverted change: status messages go on stdout, not stderr
        Added binCenter routine, which is used by the correctShiftOffset proc.
        Also re- re- re- corrected the updateUser routine, which needs to spit
        its output to stderr, rather than stdout.  It does NOTHING if xplor
        isn't called with the -verb flag, so changing it to stdout BREAKS
        the expected output.  Please leave this as is, and flag any test
        or example scripts that use -verb as errors instead.
        changed verbose output from stderr to stdout
  tcl/noe_reporting.tcl 
    Greatly expanded statistics reported by the
    summarizeMarvinPass routine. 

  tcl/noe_selecting.tcl 
    Added likelihoodLessThanOrEqualTo and likelihoodGreaterThanOrEqualTo procs
  tcl/sa_protocols.tcl 
    Added cleanup of covalent geometry XplorPots, and made some minor changes
    to the "new" pass3 protocol, which includes inverse forces and is used
    with likelihoods direct from jointFilter.tcl, rather than from a pass
    of structure calculations

  tcl/sa_tools.tcl 
    Added flag to grabPDBfiles to allow user to specify a particular
    model number within the PDB file to read

  tcl/selection_tools.tcl 
        Fixed incorrect error message

  tcl/shiftassign_tools.tcl 
    Added ConsolidateIdenticalStereopartners proc, which handles some bad
    shift tables I found on the BioMagResBank.
    
    Also added some minor bugfixes, one of which was caught by the
    initMatchTest 

Topology and Parameter files

  toppar/ion.par 
        added ion.top/par from J. Howarth - probably cribbed from CNS

  toppar/ion.top 
        added ion.top/par from J. Howarth - probably cribbed from CNS

  toppar/nag.par 
        NAG topology and parameter files contributed by John Vakonakis

  toppar/protein-1.0.par 
        added missing IMPRoper for ACE methyl

VMD interface

  vmd/publicVMDInter.cc 
        added colorMeth argument to bond method

many documentation updates


		     Changes from 2.16 to 2.17


************** C++ Layer changes **************

	now use the supplied lapack, blas by default

  CDSlib/cdsVector.cc 
	added alloc member - so that instance information can be stored.
	added loadRawArray method - to load contents from a C-style array.

  CDSlib/rc_ptr.cc
	added reset method + test

  CDSlib/sthead.hh
	added global typedefs: MutexType SemaphoreType. These can be
	overridden by platform-specific includes

  common/avePot.hh
	scale no longer sets/uses subPot's value - now they're multiplicative

  common/derivList.hh
	changed argument names so that swig would supply default Simualtion*
	arguments.

  common/ensemblePot.cc
 	sharedPID member: added mechanism to detect intervening calls
        to fork(), and fail if detected 

  common/ensembleSimulation.cc
	barrier: re-enable exception thrown upon loss of member synchronization
	moved atom data arrays out of Simulation::SharedData into its parent.
	
	Lesson: never, never place pointers in shared memory.
	mods due to changes in Semaphore, Mutex:
	  now explicitly place Simulation::bThreadMutex and
	  EnsembleMemberSimulation::sem in shared (mmapped) memory,
	  and explicitly clean them up
	
	mods due to changes in Simulation

  common/mmapAlloc.cc
	added debug statement
	alloc: added optional uniqueID argument- overrides static value
	free: now correctly handles 0 pointers

  common/mutex-sem.cc
	semaphore:
	   changed sem type to SemaphoreType - this is currently
           defined in arch.hh  (in arch/$ARCH/include) - from a
           pointer type. This avoids one memory dereference, but more
           importantly, allows better control of where in memory the
           full Semaphore object resides. 
	   [ note that previously, sem was manually placed in mmaped memory for
	     shared usage. Now a shared Semaphore should be placed in
             shared memory explicitly. ]
	
	Mutex type is now explicitly MutexType, and not a void*. The current
	implementation is in terms of an underlying semaphore. This might be a
	template argument in the future. Again, memory localization is
	simplified.

  common/pdbTool.cc
	added useChainID flag
	chainID --> segname override; initial altLoc support

  common/simulation.cc
	resizeAtomArrays method removed, replaced with resizeAtomData.
	add/clearBondedPair methods removed
	changes to support XplorWrapProc:
	  all atom, and bond data now live in member data of type AtomData
	moved atom data arrays out of Simulation::SharedData into its parent.
	
  common/xplorPot.cc
	added simulation() method
	energy and derivs methods now call down to corresponding
        XplorWrap methods 

  common/xplorSimulation.cc
        renamed static method simulation() to defaultSimulation()
	mods to accomodate XplorWrapProc
	  wrap() method is no longer const.
	  added override of resizeAtomData
	constructur: now append child pid to simulation name for nonlocal
	  XplorSimulations
	syncTo: pass modified member to slave process

  common/xplorWrap.cc
	changes as in header
	added energy and derivs methods
	made public, virtual methods to support XplorWrapProc
	added local flag
	openFile: fixed removal of ASSFIL info message
	added en/disableOutput and fastCommand specializations to XplorWrapProc
	const changes; added energy/derivs methods; select changed to virtual

  common/xplorWrapProc.cc
	XplorWrapProc: wrapper to run XPLOR as separate process, w/
        shared memory communication. This allows one to set up
        multiple XPLOR simulations with different structures within
        one instance of Xplor-NIH.

  intVar/Makefile
	added -LEXEDIR

  intVar/dint-test.cc
	fixup for printCM --> printStepDetails change

  intVar/dinternal.cc
       -printCM renamed to printStepDetails and exported to PublicIVM
       -large timestep warning now doesn't print unless printStepInfo is 
        enabled

  intVar/publicIVM.cc
	syncPos: added (untested) recentering code

  intVar/publicIVM.hh
	printCM renamed to printStepDetails and exported to PublicIVM

  nmrPot/csaPot.cc nmrPot/rdcPot1.cc nmrPot/varTensor.cc
	VarTensor:
	  added initial support for ensembleAxis()
	  this to support varTensorTools.calcTensor_ensemble

  nmrPot/noePot.cc
	added *untested* log normal potType

  nmrPot/surfaceArea.cc
	mods for getting at internal deriv info
	mods to expose class details to SWIG - mostly meaning unnesting classes

************** Platform-specific changes **************

   Darwin/Intel
	updates so that LOCAL_CC, etc work ok
	now use -llapack -lblas for LAPACK_LIB
	createF77SharedLibrary:
	  remove FLD_FLAGS
	now supply default values of CC, CXX
	added default value for F77

  arch/Darwin_8_x86/python/Makefile
	removed DARWIN_LAPACK_LIB redirection

  arch/Darwin_8_x86/source/Makefile
	disable some optimization reductions
	reduced optimization level to avoid crash in orient tests:
	  orientations, parser

  Irix
  arch/IRIX64_6.5/Makefile.defs
	added explicit def. of BLAS_LIB (to libscs)

  Linux/x86
  arch/Linux_2.2_i686/Makefile.defs
	update for intel compilers version 9.1:
	  add -axP flag which allows generation of much faster Fortran code on
	   Core machines, with a slight penalty on althons.
	for pgi build, add -lpgsse2 for lapack
	added default values for CC, CXX

************** User-invoked scripts **************

  bin/pty-xplor (used by VMD-XPLOR)
	change from an indiscriminate wait to waitpid...

  bin/pyXplor.in
	whitespace protection: added missing quotes

  bin/seq2psf.in
	added the -custom_rename option - for nucleic acid naming

  bin/xplor.in
       -now export XPLOR_EXE to environment: so XplorWrapProc can
        spawn proper exe
       -updated behavior for finish-up when running parallel jobs:
	  now wait for only the first child job (process 0). That job has
	  already waited for the other active jobs in python/xplorFinal.py

************** Documentation updates **************

  online tutorial
	merged changes from class of 2006/12/08

  online Xplor reference manual
	added references for wildcards in the NOE syntax section.

  helplib/nih-py-avePot
	scale no longer sets/uses subPot's value - now they're multiplicative

  helplib/nih-py-ensembleSimulation
	updated w/ note about straddling potential terms

  helplib/nih-py-ivm
	updated hinge() method docs

  helplib/nih-py-noePot
	updated deviation docs a bit

  helplib/nih-py-varTensor
	nmrPot

  helplib/faq/node0950.txt
	updated hyperlinks into the XPLOR manual


************** Example script updates **************

  eginput/PSF_generation
	added pdbTest  
	  download and verify psf generation and coordinate reading
	added testPDB2PSF script and support files
	  validate PSF generation.

  eginput/PSF_generation/addAtoms.py
	updated dynamics stespsize

  eginput/cmdline
	added testParallel.py
          basic validation of parallel execution and communication capability

  eginput/dna_refi/
        added ensemble.py 
          - ennsemble refinement using RDCs, CSAs, solution scattering info
	added plotIrange.py
          - calculation solution X-ray scattering curve

  eginput/dna_refi/refine_full.py
	fixup to pdbToPSF by addition of customRename arg
	move covalentMinimize to before tensor atom creation

  eginput/gb1_rdc/
	added pairRMSD.py to calc pairwise positional rmsd

  eginput/gb1_rdc/refine.py
	changed CDIH threshold back to the default value

  eginput/gb1_rdc/sa_tmv_bice_rgyr.inp
	reworked to run properly when -parallel is specified

  eginput/marvin
	removed il4, protg example cases

	Disulfides are now treated with bonds, not NOEs.

	Scripts include support for running example in quick mode.
	 quick mode:
	   -only calculates 20 structures per pass.
	   -only calculates structure of the first 36 residues of CVN.
   	   -write_subset_*.tcl does the first matching of the complete
            set of peaks to the first 36 residues.  These scripts
            write out PIPP PCK tables of the assignable peaks that are
            then used by the initMatch*tcl scripts, which should speed
            them up considerably. 
  	   -subset_talos_pred.tab is just an edited version of
            talos_pred.tab, covering only the first 36 residues.
	

	initMatch scripts use shift referencing correction, stripe correction,
	network filtering, and primary sequence distance filtering.
	
	Replaced pass1 with pass2, since I now have good starting likelihoods
	from the network filtering.
	
	resByRes.tcl produces a report of the contacts to and from
        every residue.

  eginput/marvin/cvn/check_talos_predictions.tcl
	Eliminated direct readin of parameter file--now handled
        automatically by readParamPsfPdb

  eginput/marvin/cvn/cvn_c13noe.shifts
	Corrected a wrong residue name in shift table

  eginput/protG/.runAll
	fixed calcTensor line

  eginput/protG/anneal.inp
	cosmetic change

************** Python Interface Changes **************

  python/Makefile
	dist rule: copy .lines files over
	.so build rule: added LAPACK_LIB

  python/atomAction.py
	fixes for selections entered as strings
	randomizeDomainPos:
	  various fixups
	added methods:
	  copyAtomCoords
	  randomizeDomainPos
	
	removed global import of vec3 module contents

  python/cdsList.i
	added instantiation of CDSList_double

  python/cdsMatrix.i
	added constructor from list of lists
	changed int to float to get proper type deduction of sqrt
	added typemaps for DComplex
	added SymMatrix
	added eigen(SymMatrix/CDSMatrix) functions
	added trace, det functions

  python/derivList.i
	changed argument name

  python/ensembleSimulation.i
	added fromSimulation downconverter function

  python/ivm.py
	enabled printStepDetails
	minimizeEscape:
	  changed default numSteps
	added docstring
	setBaseAtoms:
	  fix to support multiple Simulations
	breakAllBondsIn:
	  optimization: not good enough- this must be placed in the C++ layer.
	fixed treatment of finalTime when minimization==True: ignore it

  python/nonBondTools.py
	updated to use new PRINT_FILE and PREV_PRINT_FILE xplor variables

  python/protocol.py
	updated hyperlinks into the XPLOR manual
	matchInexactAtomEntry:
	  added rules for strange nucleic base HN atom naming
	revived/rejuvinated covalentMinimize-
	  it is used by two scripts in eginput/dna_refi
	initCoorde: added useChainID flag
	matchInexact: handle chainID, altLoc
	fixupCovalentGeom:
	  handle increased levels of verbosity
	fixupCovalentGeom:
	 *constraints interaction statement now placed after initNBond, so it
	  doesn't get clobbered.
	 *nicer verbose printouts
	fixupCovalentGeom: added rigidRegions arg, xSim fixup
	fixupCovalentGeom:
	  added maxViols optional argument
	  for minimization, now step dEPred to current energy / 10. This to
	  avoid large step sizes when we start very close to the minimum.
	  now print out violations in extraTerms
	updated to support multiple XplorSimulations
	multiple methods now have optional simulation argument
	removed unused covalentMinimize function
	fixups for fixupCovalentGeom
	cartesianTopology:
	  fix to allow sel object to be an AtomSel instead of only a string
	fixupCovalentGeom:
	  mods to allow this to run on nonlocal XplorSimulations:
	  use the IVM instead of XPLOR's dynamics and minimization engines
	    (note that the runtime is longer, but the number of iterations is
	     about 10x smaller)
	updated to use new PRINT_FILE and PREV_PRINT_FILE xplor variables
	initDihedrals:
	  added optional string argument, for specifying restraint
  	  table as a string 
	fixupCovalentGeom:
	  added extraTerms argument - additional pot terms to be satisfied.
	fixupCovalentGeom:
	  updated to work when using EnsembleSimulations
	initNBon: added more docstring about cutnb/tolerance relationship

  python/psfGen.py
	pdbToSeq:
	  fixes so that the segment name gets read/set correctly.
	added ACE as a protein residue type
	seqres:
	  fix case of missing chainID in SEQRES field- now we really read the
	  SEGID from the first matching ATOM field
	deal properly with chainID (override segid)
	psfGen.seqToPSF:
	  Marius' custom nucleic atom renamings are no longer longer
  	  the default but still available via the customRename flag.
	deduceSeqType:
	  added URA to URI as residue name which will trigger identification
	  as RNA.

  python/pyConvert.cc
	fromPy(double):
	  fixed error checking to not crash when a complex is given

  python/pyInterp.cc
	constructor:
	  initialize exitStatus

  python/sansPotTools.py
	create_SANSPot: added weightByExpt toggle

  python/selectTools.py
	fixes to properly deal with multiple Simulations:
	sim arguments to functions removed

  python/simulationTools.py
	StructureLoop:
	  changed default value of averageCompSel to omit axis atoms.

  python/socketComm.py
	removed debug messages
	barrier:
	  added cookie info (calling frame's line number) to barrier message
	Connection constructor:
	  handle failure of gethostbyaddr: fall back to just using ip number
	  to identify the remote host.
	Comm: constructor: increased default timeout length to 1.6 days

  python/solnScatPotTools.py
	added Chi^2 to analyze output

  python/surfaceArea.i
	mods for getting at internal deriv info
	instantiate newly unnested classes

  python/varTensorTools.py
	added calcTensor_ensemble; added support for ensembleAxis to
	saupeMatrix and analyze 

  python/xplorFinal.py
	xplor.exitStatus is now treated differently:
	  any nonzero value will cause error message to be printed, and
	  final time stats to be skipped.

  python/xplorInit.py
	parseArguments:
	  support the case that argument ``options'' is a single string
	parseArguments:
	  formatting changes in error message.

  python/xplorPot.i
	added support for restriants() method for the CDIH term.
	the method is registered during XplorPot construction

  python/xplorPotTools.py
	updates to support multiple XplorSimulations
	updated to use new PRINT_FILE and PREV_PRINT_FILE xplor variables
	added support for restriants() method for the CDIH term.
	the method is registered during XplorPot construction
	PotData constructor:
	  added support for numRestraints - for CDIH term only.
	  motivated by Jurgen Doreleijers

  python/xplorSimulation.i
	getXplorSimulation:
	  fixup so that this works for size=1 ensembleSimulations.
	moved getXplorSimulation here from protocol.py
	added converter  (downcaster) from Simulation
	added helpers:
	  enable/disableOutput
	fixups for python shell and select wrapper methods
	added python convenience methods: shell, select, command, fastCommand

************** XPLOR code changes **************

  source/ccalls.c
	getusername: fix for case of failure in getpwuid

  source/energy.f  source/etor.f   source/pick.f   source/cstran.f
	fix nasty bug in etor: the check for bond colinearity would
	fire off at fairly large values of the dot product.

  source/funct.fcm   source/selrpn.f   source/string.f
	copyst subroutine:
	  now set new copyst_err logical if target string size is too small
	selrpn:
	  make use of copyst_err to set selerr appropriately.
	
	This allows C++ AtomSel object to throw an exception on such errors.

  source/hbuild.f
	format statement syntax fixup

  source/noe.f
	added log-normal potential type

  source/parmio.f
	fix for problem first noted in 1998:
	  http://www.bio.net/bionet/mm/x-plor/1998-August/001729.html
	
	vdw radii were incorrectly calculated from nbfix statements. This
	impacts the repel potential, and allows atoms to get too close.

  source/parser.f
  source/xplorFunc.f
	added DECLARed variables PRINT_FILE and PREV_PRINT_FILE
	involved increasing the length of CMDPMX in symbol.fcm
	getvar_asString:
	  now calls TRIMM on returned string

  source/prot.f
	Correction to the electric field part of proton shift
	calculation.  I wasn't summing up the effects of every heavy
	atom on each proton--rather, I was just returning the last
	heavyatom's effect on each proton.  David Wilton from 
	Mike Williamson's lab caught this.

  source/xplor.c
	added -sharedProc cmdline flag with call to C++ xplorwrapproc_slave
	changed new_argv initialization to ANSI C - for pgcc

  fortlib/lapack/Makefile
	no longer include routines which are already in blas

  fortlib/lapack/dsecnd.f
	commented out unnecessary external declaration for ETIME intrinsic

************** Changes to the Marvin/PASD facility  **************

  marvin/MarvinNOEPotential.cc
	-Eliminated generation of inverse exceptions for peaks expected
         to be near the diagonal, since that's handled by TCL.
	-Also eliminated a double-counting of accounted & unaccounted
         neighbors in updateNoeCompleteness.  The original code
         produced the same numbers as the current code, but spent twice
         as long doing it. 
	-Added method to reset the explicit inverse exceptions list

  tcl/aeneas_tools.tcl
	Added initializeLikelihoodsFromFilters
	Several changes:
	1.  match* routines now check whether a ShiftAssignment
            actually has a proton/heavyatom shift before trying to
            match it to a peak.  To support an (unused by default)
            option in stripe matching, which turns off SAs that don't
            have stripe targets. 
	2.  Eliminated recenterShiftAssignments.  Replaced with
            stripeCorrection. 
	3.  Eliminated origGenerateNetworkLikelihoods,
            try2GenerateNetworkLikelihoods, maxLinksInAgreement, and
            generateNetworkLikelihoods.  Replaced with networkFilter. 
	4.  NetworkFilter can accept multiple pots and analyze them together.
	5.  Eliminated stripeFilter and gatherTargetShiftValues.
            Replaced by stripeCorrection. 
	6.  Added correctShiftReferencing.  Detects any constant
            offset between the from and to proton dimensions and
            corrects peak positions to eliminate it. 
	7.  Added ability to create (with createExplicitExceptions)
            explicit exceptions for peaks expected to appear within solvent
            lines. 
	
	Note:  file also includes beginnings of secondary structure
         Hbond detection (with secStructFilter), but it doesn't work
         well yet, and is not used in released scripts. 

	addition in stripe filter--ability to update shiftAssignments' 
	shift values to the targets'.
	
	Lots of changes to the netFilter code, but mostly to the
        obsolete versions. 
	Modified enforceExpectedSymmetry proc into new symmetryFilter proc.
	
	match*d now uses improved mechanism for handling unfolded
	peak locations.  UnfoldPeakPosnToMatch is now the key proc,
        and it always returns two possible unfolded locations:  one
	with an even number of folds, and one with an odd number of
        folds.  match*d tries all combinations of those possibilities 
	in searching for peak position/unfolded SA posn matches.

  multiple files:
	Update to use atomSel class's == and intersects
        operators. Replaced occurrences of old TCL procs atomSelsEqual
        and atomSelsOverlap. 

  tcl/import_pipp.tcl
       -Added a proc to write PIPP PCK tables.  Doesn't yet handle PIPP-style
	assignments.  Also doesn't handle writing slice numbers.  Also missing
	ability to generate the DATA AXIS lines that are used to set spectral
	ranges.  Other than that, it's quite complete :-/
       -Improved comment-skipping in PIPP tables
       -Added rules in code for importing PIPP shift tables, ignoring
        lines beginning CUR_TYPE and ignoring lines commented out with
        a leading #, as Dan Garrett sometimes has.

  tcl/noe_reporting.tcl
	Changed format of one output line.
	Several additions:
	 1.  reportNOEaccuracy can now report on the accuracy of
	 peakAssigns above a certain likelihood, or those below a certain
	 likelihood.  Used to eval accuracy of accepted peakAssigns
	 and rejected peakAssigns, respectively. 
	 2.  reportNOEprecision now also reports on NOE
 	 discrimination, the fraction of peakAssigns that is either >
 	 a high likelihood cutoff or < a low likelihood cutoff.  High
 	 NOE discrimination correlates with convergence. 
	 3.  markGoodPeakAssignments reports number of unassigned
	 peaks, for convenience. 
	 4.  initialPeakAnalysis can now be set to report statistics
	 based on counting all peakAssigns, or only those whose
	 likelihoods are within a given range. 
	 Useful for evaluating performance of filters.

	-Added calculation of fbad-long using previousLikelihoods
	 arising from peakAssign filters 
	-Expanded residue-by-residue reporting function to combine
	 several pots' peaks into one report.
	-Modified summarizeMarvinResults proc.  Added separate flags
	 for violation and high-likelihood values to be used while
	 rescuing filtered PAs, so that the structures' accuracy need
	 not be too great to bring in filtered PAs. 
	-Added logic in PA rescuing, so that I don't add PAs to peaks
	 that already have high-likelihood PAs.  
	-Expanded reporting of the filtered-PA-rescuing stuff.

  tcl/noe_tools.tcl
       -Corrected wording of a thrown error--old version indicated
	wrong flag syntax 
       -A few Marvin peak reading & writing improvements:
       -Extended ability to parse both current marvin-formatted .peaks
        files and older marvin-formatted .restraints files.
       -Improved xplor-formatted output.  Added check for
	peakAssignments with intersecting to & from selections, so
	that they won't be written to xplor files--classic xplor can't
	deal with them.  
       -Also improved handling of bounds corrections--I now find the
        maximal upper and lower bounds corrections for all the
	peakAssignments of a given peak, and use those in generating
	the xplor ASSIgn's distance bounds. 
       -added support for Michael Nilges's extended-OR format for NOE
        ASSIgns.  If a given marvin peak has multiple peakAssignments
	to be written, the peak is written as an 
	ASSIgn (sel) (sel) d dmin dplus OR (sel) (sel) OR ... 
	restraint, which the current classic NOE code can parse correctly.
       -Two peak-reading bugfixes, plus I moved removeLowLikelihoodPAs from
	aeneas to here, and added removeFilteredPAs.  RemoveFilteredPAs is used
	at the end of initMatch scripts to clear out the filtered PAs.
       -RemoveLowLikelihoodPAs is used in convertToXplor, as a final
        cleanup before creating xplor-style NOE tables.
       -Added consolidateClosePeaks (from make_random_noes.tcl).  
       -added support for reading & writing the number of failed
        filters for each peakAssignment.  
       -Caught a few bugs in the xplor-format NOE reading routine.
       -now remove all explicit inverse exceptions before reading a
        new file of them 

  tcl/pdb_support.tcl
	Added support for reading specific models from a PDB file

  tcl/sa_protocols.tcl
	pass? protocols have been reworked.

  tcl/sa_tools.tcl
       -added code to print a Python traceback when an exception is encountered
       -Eliminated minor memory leak from non-cleaned-up atomSel.
       -Includes beginning of PDB cluster analysis.  Not used in
	released scripts. 
       -use appropriate env variables instead of paths relative to XPLOR_DIR

  tcl/selection_tools.tcl
       -Added a fairly complete test script for all the procs in
	selection tools.   
       -Some bugfixes to selection tools, engendered by the test.  
       -new code: previous and nextResidue procs
       -memory leak fixes in makeSelectionNonStereospecific and in
        stereopartnerForSelection. 

  tcl/shiftassign_tools.tcl
       -Added ability of initialShiftAssignmentAnalysis to make
	reports based on all peakAssignments, or only on those with
	likelihoods within a given range. Analagous to the changes in
	initialPeakAssignmentAnalysis. 
       -Corrected & expanded reporting of NOE completeness, so
	that it agrees with numbers generated by 
	MarvinNOEPotential::updateNoeCompleteness.  Also reports 
	specifically on accuracy of explicit exceptions that do not
	correspond to a peakAssignment (ie., were generated by network
	filtering, or diagonal exclusion) 
       -added proc to group shiftAssignments by residue, which is used
        in a few places (eg., netFilter's intraresidue neighborhoods)

  tcl/talos_tools.tcl
       -Added proc to read in a TALOS .tab file and write out an xplor
	cdih table. Previous code reads TALOS and creates dihedral
	restraints directly, with no chance to output an
	xplor-formatted file.  This isn't pretty, but it'll do 
	for now.  Useful when going to python scripts, where TCL
	readTalos isn't convenient.
       -Added ability to skip commented lines

************** Parameter/Topology updates **************

  toppar/nucleic-1.1.par toppar/nucleic-1.1.top
	intended changes somehow got dropped from the previous version
	version string in note updated
	added missing impropers for NH2 base groups and H2'/''
	bumped minor version number
	fixes to stereochem defs to match IUPAC - contributed by J. Doreliejers

  toppar/nucleic-2.0.par toppar/nucleic-2.0.top
	naming corrections
	added missing impropers
	version number bump to 2.0

  toppar/protein-2.0.par toppar/protein-2.0.top
	naming corrections
	added missing impropers
	version number bump to 2.0

************** VMD interface changes **************

  vmd/vmdInter.cc
	remove some ill-advised radius-setting statements


------------------------------------------------------------
		     Changes from 2.15 to 2.16

General 

 now provide native intel/Mac binary package

 documentation:
   XPLOR reference manual has significant updates
   various updates/corrections to Python reference manual

 mailing list now searchable from 
   http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih -- thanks to Gmane.

 pty-xplor:
  added the -xhost command-line option
 
 xplor, pyXplor scripts:
   now protect space, special characters in command-line arguments

XPLOR interface

 added pseudo-CSA (PCSA) potential term, test case from A. Grishaev
  see A. Grishaev, J. Ying, A. Bax (2006) "Pseudo-CSA restraints for
  NMR refinement of nucleic acid structure" J. Am. Chem. Soc. 128,
  10010-10011. 
 
 
 source/ccalls.c
  frename: now remove trailing spaces from old as well as new
    (contributed by J. Howarth)
  frtn_gethostname: workaround for some broken gethostname
  added new FRTN_CLOCK c function
 
 source/cstran.f,v
  changed dihe restrain info message, and now print to DUNIT
 
 source/fantalin.f
 source/fantacross.f
  output fixups, inspired by J. Howarth
 
 source/hbdb.f
  fixed array bounds error noted by J. Howarth
 
 source/orientations.f
  added ugly hack/memory leak to get working on Darwin_x86
 
 source/saxs.f
  saxsinit2:
    initialize glocc array to 0
    rearrange argument decl's, variable definitions to normal conventions
 
 source/linuxtime.f
 source/x86_64time.f
  change VCPU to call new FRTN_CLOCK c function
 
 source/xdipolar_coup.f,v
  fixed rhomicity definition in help text
 
 source/xplor.c
  changed argument treatment: now all options we know about are
  removed from the arg list (except for -tcl)
 
Python Interface

 builtin xplor module
  added helper function xplor.parseArguments() for argument parsing and 
  validation. 
 
 varTensor:
   will now throw exception if Rh (rhombicity) is specified outside the legal 
   range
 
 cdsVector:
   added constructor taking size and fixed value arguments
 
 findPython:
   USE_MACOSX_FRAMEWORKS is now an env var
 
 ivm.py:
  subtle change to breakBond: in 1st arg passed to convertToSelectionIndices
 
 protocol.py:
  matchInexactAtomEntry:
    -added rule matching H to HN
    -added matching rule, examining final character. Came from ILE
     difference in XPLOR, charmm22 ff.
  fixupCovalentGeom:
    -change treatment of sel argument: do not re-evaluate selection
    -save and reset mass values
    -placed default tolerances in argument list
    -split constraints interaction selection line so that longer
     selections are possible before COMAX errors.
    -exit: fixup exit message, and customize the exception thrown
  initCoords: 
    added optional erase argument
 
 psfGen.py:
  dna residueType: added URI residue
  deduceSeqType: 
    differentiate DNA/RNA by presence of URI
  pdbToSeq:
    fixed extraction of the resid field to get all four characters.
  pdbToPSF: 
    changed seqToPSF to not deprotonateHIS- can be done in pdb
 
 pyConvert.hh:
   added toPy(FloatPair)
 
 pyXplor.cc:
  pythonStartup:
    now employ xplorInit.py to do initial argv parsing
  python_interp:
    now executes xplor.scriptName (instead of first item on argv),
    if it's not empty.
  change:
    now run xplorInit.py in the context of the module ``xplor.''
  xplorInit:
    much less is now enclosed in strings
    added parseArguments helper
 
 sansPotTools.py:
  exchangableName: 
    now print warning message if residue is not supported, don't crash
  moved solventVolumeSet definitions to solnScatPotTools
  create_SANSPot:
    added preweighted option
    added radiusScale (now properly treat excluded volume)
    added fractionD2O (now deal with D2O solvent)
    added exchangeFraction (to deal with exchangable protons in D2O)
    added normalizeIndex argument
    now clip negative I values to zero
  added useGlobs function and glob defs.
   -support separate proton globbing
  changed scattering length units by 10
    also fixed units in comments.
  changed used SolnScatPot version from complex to real
  added specialized useGlobs frontend
 
 simulationTools.py:
  genFilename:
    updated to use xplor.scriptName.
  StructureLoop:
    added averageSortPots argument - terms to use to sort structures
    genAveStats: 
      now call averageContext inside stats loop
    added averageTopNum as alternative to averageTopFraction
    added averageAccept function argument to report on only allowed structures.
  StructureLoop.genAveStats:
    deal properly with the case of no acceptable structures
    change location of averageContext call: now done for each structure,
    so that things like globCorrect() get set properly.
  added convertToPotList function
  testGradient: added alwaysPrint option
  documentation update
 
 socketComm.py
  startServer: write message to console on failure
 
 python/solnScatPot.i
  added pairDistribution function
  added import during module initialization
 
 python/solnScatPotTools.py
  add Pr function:
    weighted pairwise distance distribution function.
    supports EnsembleSimulations with proper averaging.
  useGlobs:
    added verbose entry which turns on single-glob atom listing
 
 python/solnXRayPotTools.py
  useGlobs: support for verbose argument
  moved solventVolumeSet definitions to solnScatPotTools
  moved useGlobs from solnXRayPotTools.py to solnScatPotTools.py and
    generalized, then placed wrapper back in solnXRayPotTools.py
 
 python/varTensorTools.py
  calcTensor:
    now able to calculate the tensor from an ensemble of structures (not
    necessarily an EnsembleSimulation)
 
 python/xplorInit.py
  fixed up to deal with command-line options, arguments
  added xplor.parseArguments:
    uses writeConsole to print command-line error message
  change:
    now run xplorInit.py in the context of the module ``xplor.''
    much less is now enclosed in strings
 
 eginput/gb1_rdc:
   .py files updated
   StructureLoop in .py files now implements averageSortPots argument
 
 eginput/protG:
   added calcTensor.py to calculate alignment tensor from expt rdcs + structure
   grid_search.inp - add comments on use
   tang.py - updated to deal with segids, unknown resid types
 

TCL interface

 tcl/Makefile
  deal with empty TK_INIT_DIR def
 
 tcl/findTCL
  handle case of missing tk_library
  USE_MACOSX_FRAMEWORKS is now an env var
 
C++ Layer Changes

 rdcWavePot:
   updated
 
 atomSel.cc:
   reimplemented intersects() method for speed.
 
 noePot.cc:
   updated defaults to be consistent with those in noePotTools
   change: now default to HARD potType
   NOE tables
     increased maximum distance to 10^6 (in ASSIgn statement)
     added a check that atom selections don't intersect.
     allow zero distances
 
  


		     Changes from 2.14 to 2.15

General

 * Parallel job synchronization now uses tcp communication
     - allows some nodes to be down at startup (will ignore those
       hosts). Also allows for xplor-nih or whole machines to crash
       and not hang the parallel calculation- structures scheduled on
       those nodes will be ignored. 
     - requires communication over port 9021 (or that specified by the
       -pport option)


 *Changes to the command-line xplor script:

  -fixed -smp so that -parallel need not be specified.

  -support for new socket communication facility:

     added support for socketComm functionality with -pport, -phost options
     and XPLOR_PPORT, XPLOR_PHOST env vars.

     added -crash_on_startup flag to cause crash if a process doesn't
     report in at startup

     (for parallel jobs) now redirect file descriptor 9 to stdout
     before redirecting to the log file so that we're able to send
     messages to the user: parallel calcs now print cpu info summary 
     to stdout at exit


  -remove LD_ASSUME_KERNEL hack - this caused problems on newer
   distributions. This change may break things on FC2.


 *startup message: now include the hostname

 *Parameter changes:

   -protein-1.0.par:
     added missing angle parameter for SP patch
   
   -protein-1.0.top:
     fixed the YP patch 1cz 1oh 2p angle definition
  

Python Interface

  Binaries now built with Python version 2.4. Source also builds with
    version 2.3

 ./configure: now always performs Python precompilation to avoid race
   condition on parallel startup. Also precompile files in bin.$ARCH
   Fix XPLOR scripts so that sure we're always using optimization.

  atomSelAction: Fit: 
    fixed ugly bug which could cause the rotation matrix to become an
    inversion matrix.
  
  most C++ potential terms with restraints:
    added numRestraints() method
    added showAllRestraints boolean - if set all restraints are output
    in the showViolations() method (but violated terms are flagged).
  
  PDBTool: added support for PDB MODEL records
    read() now returns list of unmatched PDB ATOM entries
  
  PotList: 
    change to rms(): now use individual terms' scale() to weight
                     the rmsd calc, if the terms are all of the same PotType.
                     also weight by # restraints.
    Also made a fix to violations().
  
  Simulation::atomString:
    output has been rationalized.
  
  SimulationWorld:
    added cpuTime method
  
  XplorPot:
    added delegate so that rms(), violations() methods return useful values
  
  XplorWrap:
    select() will now raise an exception if there is an error.
    added fastCommand() method
  
  
  PREPot:
    update adding SUM restraints - PRE contributions from
    distinguishable groups which contribute to the same relaxation process
  
  SolnXRayPot: renamed to SolnScatPot. 
    Now supports complex scattering amplitudes
    added glob defs for DNA.
  
  cdsMatrix:
    fixes to *, - operators
  
  cdsVector:
    fixes to - operator
  
  ensembleSimulation:
    added helpers: singleThread, multiThread, commBarrier
    better EnsembleSimulation cleanup at exit
    Modified behavior with EnsembleSimulation fixed
    
  protocol.py:
    initCoords: 
      added support for optional model parameter
      added support for unmatched pdb ATOM entries. guessing logic-
        togther with the with matchInexactCoords helper.
    initDihedrals: fixed default chi2 restraint definitions.
    initTopology: fix to allow pathnames relative to pwd and TOPPAR (and absolute
                  names too).
    added initRandomSeed, initialRandomSeed helpers
       Distribute seed value to other parallel processes if the seed if
       not explicitly specified. 
  
    torsion/cartesianTopology:
      added oTensors argument which takes a list of VarTensors to set up
      according to the setFreedom() method.
  
  psfGen:
    cisPeptide: fixed so that it actually works for the non-proline case.
    seq2PSF: fix in several patchings to deal properly w/ multiple segids.
  
  pyInterp:
   allow triple-single quotes in scripts redirected from stdin.
   Python shell: better handling of error conditions- exceptions are now 
   thrown back to C++, proper exit status is returned
  
  python startup:
    errors in startup will now throw exception. Previously only a
    warning message was printed.
  
  rdcPotTools:
    fixes to dipolar coupling scaling.
    Allow using correct gyromagnetic ratio signs.
    added DaScale helper function
  
  restraintStats:
    fixes for violation amounts
    now accumulate and print crossTerms separately from main potList
  
  simulationTools:
    StructureLoop: 
      added averageCrossTerms argument to constructor - so that
        non-refined-against terms are listed in statistics, average
        printouts
  
      allow averagePotList, crossTerms to be plain python sequences
  
      added new optional paramter: structureNums. Can use
        this instead of numStructures, startStructure
  
      calcAverageStructure:
       now treat ensembles correctly: calculate the average of the
         ensemble-averaged structure.
  
      averageCompSel changed from "not hydro" to "not name H*"
      genAveStats: now print out the rmsd values for fitted selection in
                  addition to the comparison selection. Also write out
                  these two strings.
      writeStructure: 
        added extraRemarks argument
        now writes out seed from protocol.initialRandomSeed
   
     now report pos. rmsd to ave. of calcd struts

  analyze():
    can now give a single Pot as an argument (doesn't have to be a sequence)

 varTensorTools:
   calcTensor:
     no longer reset oAtom to origin- rather, keep it at its current
     position.
     fixed handling of non-NH normalized restraints
     fixed handling of restraints w/ more than one bond vector
   analyze:
     fixed calculation of Euler angles
   added xyzFromEuler helper
   configIVM_fixAxis/RhToOther:
     fixes for ensemble simulations.

 vmdInter.py:
    VMDObject.labels: now allow sel arg to be string
    change to object naming so that using file in subdir works

 xplorPotTools: now supports COLL term

 added minimize.py - optimization algorithms


 various small updates to the examples
   gb3_ensemble: added order.py example script - refine against order
   parameters 


Platform Specific

 *Darwin: 
   -threads now supported in VMD interface
   -updated to be compilable by gfortran
   -rationalized Python libraries linking- they no longer need to be
    linked to the xplor binary 

 *better support for running on Scyld clusters:
    -can now specify SCYLD_PROCESSES env var and command-line -parallel
    -now eval $cmd so that redirection works on scyld clusters

XPLOR Interface

  *enbond: moved NCSOP to heap, removed MNCSYM parameter
  
  *hbdb: removed naamax parameter, and now allow it to grow dynamically.
         fixed an array range underflow in hbdbpatt
    
  *plane: PLNINR: reinitialize PLN_NUM to 0 so that the init statement works.
  
  *xplor.f: change behavior for case in which a filename is specified:
    now redirect it into stdin, so that the PYTHon and TCL XPLOR
    commands work in this mode. 

  *new variables are defined: START_TIME, START_DATE variables
  
  *tutorial/nmr_ensemble:
    filled in missing contents

  *multiple modifications to modernize the code - thanks to J. Howarth.



		     Changes from 2.13 to 2.14


XPLOR interface changes

  new XPLOR SAXS module from Alex Grishaev
     A. Grishaev, J. Wu, J. Trewhella, and A. Bax, JACS 127, 16621 (2005)

  ncs.fcm:
    removed MAXNGR static arrays - now dynamically allocated
  
  hbdb.f:
    removal of pause statements, prompted by J. Howarth
  
  XPLOR tests:
    added new sani and saxs tests
  

Python Interface changes

  new solnXRayPot module for refinement of SAXS and large-angle xray scattering
  data.
  
  improved support for Potential terms derived from C++ terms. Eliminates 
  the step of reconstituting a Python wrapped object by deducing its type.
  
  XplorPot:
    threshold, rms(), and violations() methods now work for supported terms.
    simulationTools.xplorThreshold has been removed.
  
  XplorWrap:
    added new methods: disableOutput, enableOutput to manipulate
      XPLOR's echo and wrnlev variables
    for convenience, xplorInit places references to these in the global xplor
    module.
  
  IVM:
    fix for error condition in buildNode:
      better identify bend atoms in bad hinge specification
      null out IVM's tree_ member when throwing an exception -
         so that it doesn't get destructed in this state (causes crash)
    
  python cdsMatrix module:
    added CDSMatrix_complex
  
  cdsVector: 
    added dot(), __mul__(float), __pow__(), sum()
  
  analyze functions: now work better with AvePot terms
  
  noePotTools:
    create_NOEPot: now initialize aveType to "sum"
  
  nonBondTools.py:
    made messages from XPLOR interface a bit less verbose
  
  rdcPotTools:
    specified different nucleic acid NH distance to use in calcing 
    Da_prefactor.
  
  protocol module:
    initNBond:
      added selStr argument. onlyCA is now deprecated.
      omit ANI residues in constraints interaction statements
    initDynamics: 
      added parameters eTol_minimum, etol_factor
    added addDisulfideBond helper function
    torsionTopology: modified to call selectTools.breakDisulfides()
  
  structureLoop module:
    StructureLoop:
      updated so that calcing ave. struct. works for EnsembleSimulations
      when generating violation statistics, now only use averageTopFraction
        structures (not all structures)
      disable creating backups for fit structures if averaging or taking stats
    StructureLoop.pdbFile():
      now use old object if the name has not changed. This necessary to
      retain remarks info.
    better per-restraint statistics over all calculated structures.
  
  varTensorTools:
    fixes for fixAxisToOther function
    added orthogonalize() helper function
  
  Python Swig change for C++ ACCESSOR and MODACCESSOR modules resulting
   in large size decreases, simplification in coding.

  added posSymmPot: position symmetry- a more powerful version of the NCS term.

  Python tests:
    added rdcPotTest3, mat3Test, solnXRayTest

  Python docs: 
    csaPot: fixed def. of csa tensor atom s in assignment table.
    mat3: updated SymMat3 constructor, element access member docs
  
  
TCL Interface changes

  PASD/Marvin:
    renamed class Restraint to Peak.
  
  tclPosnSymmPot:
    Support for Atom and AtomSel based accessors

Changes to the eginput example scripts:

    PSF generation: added complex circular peptide example

    dna_ref: added refine_full.py script

    gb1_rdc: added standalone analyze.py script
             anneal.py: changed noe setup so extra terms can be added easily 
                        updated for XplorPot threshold change
    protG: anneal.py: updated to produce .viols, stats files
           anneal.inp: removed spurious NOE parameter clause 
                       (thanks to Francesca Marassi)
  


Parameter/Topology changes
  
  nucleic-1.0.par
  changed the 5-terminus H'-H'' improper angle values to match it's
  minimized value
  
  protein-1.0.top
  phosphorylating patches: fixed patch SP, added patch YP
  
  
low-level changes:

  CDSlib/complex.hh:
    added operator-=
    added operator*(Complex,T)
    added function IEXP(floatType)
  
  CDSlib/mnbrak.cc:
    small fix
  
  vec3.hh:
    added explicit constructor from integer, so the literal ``0'' can be used.

  modified.cc:
    registerTo: don't register sim if it's already been registered.
  
  avePot.cc/hh:
    fixed calcEnergy/Derivs s.t. the scale factor is not applied twice.
  
    overide scale, threshold accessor methods to use values from the
    subpot. Note that the wrong value will be obtained  if the object is
    accessed as a Pot (the base class). Virtual methods might be a better idea.
  
  Fortran code:
    removal of arithmetic if statements, (courtesy of J. Howarth)
  
  prePot:
    optimization in inner loop

  build change:
    now should better handle transitive linked shared libraries
  
  now include required lapack, blas source- but it's use is optional.

  mac build now on Darwin_8, (OS X 10.4.4) no longer tested on Darwin_7

  added gamma and rjbesel functions from the specfun netlib library.



		     Changes from 2.12 to 2.13


PASD/Marvin:
  - fixed so that the eginput scripts now work.
  - changed linear pot to have continuous derivatives
  - improved error checking.

eginput scripts:
  added gb1_rdc/violStats.py - script to gather violation statistics
      over multiple structures.
  updated gb1_rdc/anneal.py to collect violation statistics

  updated runAll script so can now specify multiple target directories.

XplorPot:
  a single instance can now specify multiple XPLOR terms. This to work
  around errors when VDW and PVDW are called separately.

noePot:
  added Average PotType

updated Python interface tutorial

Source/lowlevel changes

  created fortlib directory and moved libfft there
  added pppack to fortlib
  template library: added Spline class 

  C++ cleanups. Contribution from Jack Howarth.

  EnsemblePot now derives from ModifiedBase
  
  Pot:
    now derives from ModifiedBase
    added threshold MODACCESSOR.
  
  EnsembleSimulation:
    fixed up addDependent, removeDependent, markAsModified methods
  
  Pair class is now a template in CDSlib.

  XPLOR interface:
    updated sbound.f to remove NMCTAB parameter
    nbonds.f: don't print close atom warning messages if messages=off.
  
  bin/makedepend:
    now omit entries which end with a slash (/) so it doesn't create
    dependencies on directories.
  
  testDist:
    now remove WORKDIR before running tests
  
  noePot, rdcPot, csaPot, jCoupPot:
    now have direct access to list of Restraint objects 
    added violation value to printViolations() header
  
  
  Python interface:
    added spline module
  
  Python Interface:
    fixed violation reporting in *Tools.py:
     threshold values are no longer reset to 0 before analysis.
  
  monteCarlo.py:
    bug fix for randomizeTorsions (affected Marvin/PASD)
  
  split off nonBondTools.py from simulationTools.py
  
  protocol.py:
    contributed by Greg Zornetzer:
    fix for fixupCovalentGeom():
      specifying a selection with bonds to a non-selected region would
      mess up the interface region.
    addUnknownAtoms():
      added arguments: dyn_stepsize, dyn_numStepMull, verbose
      reduced default stepsize to 0.02ps
  
  simulationTools.py:
    StructureLoop: 
    added genViolationStats flag to constructor, for printing out
    violation statistics.
    if averaging structures, overwrite the calculated structures with
    versions fit to the first structure.


		     Changes from 2.11.2 to 2.12

Frontend Scripts

 - seq2psf: 
    added -amidate_cterm option
    multiple chains are now supported
  
 - xplor script:
    more relaiable treatment of the PYTHONHOME env variable


Parameters

 -  added fixed nucleic.top/par topology/parameter files.


IVM

 - fixed bug for torsion angle dynamics with a bond which is initially 
   *exactly* parallel to the xz plane.


Python Interface

  - new functionality in protocol.py:
    addUnknownAtoms() see example in eginput/PSF_generation/addAtoms.py 
  
  - simulationTools.py:
    StructureLoop:
      added writeStructure() method which creates a .viols file for each structure

  - testGradient:
    fixed so that it works with an non-ensemble Simulation
    
  - changed PotName for some terms to be more consistent
  
  - atomAction.py:
    SetProperty: fix so that properties with Capitalized letters (such as
      residueName) are handled properly.
  
  - cdsVector.py:
    norm: now always reduces to a float_type scalar return value
    added vec_norm(): performs a norm on each element of a vector
    added log()
  
  - simulationTools.py:
    analyze(): terms now register themselves for reporting using 
               registerTerm(), and simulationTools doesn't need to know 
               about them.
    fixes for nonbonded analysis for repel=off
  
    added fileSuffix global variable
    used when creating temporary files to avoid races in EnsembleSimulation
    calculations
  
  - simulationWorld.py:
    added pure-python world() function to access the simulationWorld()
    static function.
  
  - trace.py:
    added function addModuleName() to make tracing arbitrary modules easier.
 
  
  - protocol.py:
    fixupCovalentGeom: allow sel argument to be a plain string
  
    initDynamics(), initMinimize(): 
       no longer set ivm's constrainLengths. This is part of the topology setup.
  
    added cartesianTopology() helper for ivm setup.
  
    added initTopology()
    initParams/initTopology(): will now take sequence type specifiers which
      use the new parameters, topology files configured in module local 
      dictionaries.
  
  - psfGen.py:
    now add remark to autogenerated PSFs - to denote their origin 
    * seqToPSF:
      now use protocol.initParams/Topology for initialization.
      added the amidate_cterm option
      changed default seqType to 'auto'
      added disulfide_bonds/bridges arguments.
      added support for seqType 'rna'
    * pdbToPSF:
      now use seqType 'auto' in call to seqToPSF
  
  - pyPot:
    rewrote the PyPot constructor to remove need to supply the
    second self argument.

  - added tests: tkInterTest, ivm3, bigPatch
    note that tkInterTest will fail if DISPLAY is not set (or X11 access is 
    otherwise not available)  .


XPLOR interface

  - optional output reduction improvements
  
  - updates to allow extra-large residues in patching (removed SEGDIM 
      parameter from segmnt.f) 
  
  - fixes to problems in Fortran code discovered by Darwin's GuardMalloc
  
  - fixes to some paraRestraints code
  
  - fixes to machine dependent code found by gfortran developers
  
  - fixes to deal correctly with case (>8 bonds)
  
  - torsion angle database term: fixed bugs caught by Guillermo Bermejo

VMD-XPLOR interface

  - fixed race condition which caused errors when trying to follow the 
    dynamics of an ensemble.

  - pty-xplor: set pager to cat- real pagers cause pty-xplor to hang in 
    python mode

  - added findXcookie to bin files distributed - for more reliable remote 
    Xplor-NIH/ VMD-XPLOR execution.

Low-Level Changes

 - atomSel:
   added intersects() operator==() methods,

 - template libraries:
   cdsIterator.hh: mods so that can be re-assigned.
   cdsString.cc: added copy() method which returns a non-reference counted copy.
   cdsVector.hh: added scale() method for symmetry with CDSMatrix class

 - compilation/development
   added support for intel Fortran/C++ compilers, version 9

   added version 1.3.25 to list supported swig versions

   findPython: now also defines PYTHON_VERSION

PASD/Marvin:

  - Significant changes to the Marvin/PASD code -
     *** old scripts will need to be modified ***

  - note carefully the following change to the initial matching script:
       now, the spectral range must be specified by giving the actual
       beginning and end of the spectrum (in ppm) along each dimension.

  - introduce new datatype, ShiftAssignment 
    * represents a single entry (or pair of covalently bonded entries)   
      in a chemical shift table.  
    * facilitates analysis of the likelihood of a particular spin's
      location in the structure and/or the likelihood of its chemical
      shifts being correct.  
    * note difference from the previous version, which only kept track
      of the likelihood of a particular assignment of a particular NOE
      peak being correct. 

  - full discussion of these changes will be available in an upcoming
    paper.




		     Changes from 2.11.0 to 2.11.2

PASD/Marvin fixes to correctly handle symmetry peak filter.

Platform specific updates:
  - Darwin version 8 (Tiger) is now supported
  - amd64: added some support for use of Pathscale fortran

frontend xplor script
  - consistent use of TMPDIR env variable, if it is set. 
  - semaphore cleanup:
    grep for current user- to avoid permission violation messages.

Python Interface:

  protocol.initNBond():
    added onlyCA flag to enable only CA-CA repulsion.
  
  gb1_rdc/refine.py example script:
    - further improvements s.t. no explicit parameter configuration is
      performed in the structure loop.
    - now will compute average structure.
    - increased the rdc force constant
  
  simulationTools.analyze():
    - now report the number of vdw violations
    - now print violation info in a .viols file.
  
  simulationTools:
    - added waitForProcess() barrier function.
    - StructreLoop:
      * added capability to compute average of all structures
      * added analyze method which prints violation info to a
        file with the structure filename + a '.viols' suffix.
      * added StaticRamp class
      * added InitialParams, FinalParams class, which initialize or
        finalize ramped parameters.
  
  fix to CDSMatrix::resize to copy old contents
  
  added CDSVector_int to cdsVector.py

  CDSString: optimization of operator+=(CHAR)
  
  PotList:
    added a copy() method

  EnsembleSimulation input multiplexer (using stdin) - now only use if
  *both* stdin and stdout are ttys.


XPLOR source:
  - removed/converted blanket SAVE statements in darwintime.f and dynio.f
  - ccalls.c: frtnchdir:
    reworked to use Fortran's string length argument.
  - update from A. Grishaev - updates the segm statement.
  - nbonds.f:
    * added DECLAR statements for REPEL nbond parameters
    * removed limit on iatmx (max num of bonded atoms) by overloading the
      CMPLTD return code of MAKINB.
  - combined xlftime.f and darwintime.f --> darwintime.f
  - noe.f:
    reduced the severity of error of calling PRINT for classes in which 
    POTEntial is set to SYMM, HIGH, or 3DPO.

parameters:
  added patches, parameters for phosphoserine, phosphocysteine to the
  standard protein.top/par files.

  

		     Changes from 2.10 to 2.11

- xplor run script: 
  added the -scyld cmd-line argument for running on scyld clusters.
  added the PROTSHIFTS env var for access to proton chemical shift database.
  under Linux, now cleanup unused Semaphores at startup.

low-level:
  - fixed error in reference-counted classes which rendered them
  non-thread safe.  
  - fixed Solaris-specific ensemble bugs.
  - fix for strange Solaris /bin/sh behavior

 - pty-xplor (used by Xplor-NIH) fixups under Solaris

Python interface
 
 - now correctly build and test Tkinter package.

 - pyXplor script: now properly protect spaces in command-line arguments.

 - psfGen module:
   added the dAmino() function to convert a residue to a D-amino acid.

 - simulationTools:
   now perform NOE analysis.

 - added NOEPotTools- helper functions including create_NOEPot()

 - protocol.py: 
   initMinimizie: added added dEPred to argument list.
   added initCoords() helper function.

 - varTensorTools:
   added calcTensorOrientation function

 - xplor (toplevel) module:
   added the requireVersion() function

 - added posRMSDPot potRMSDPotTools
   for both RAP and bfactor refinement.


XPLOR interface:

 - hbdb: fixes for problems with long filenames.

 - more robust determination of username.

 - Darwin:
   fix so that file versioning works properly.


Examples in the eginput subdirectory:

 - added gb1_rdc simple examples using NOEs and multiple
   RDCs. Included are both XPLOR and Python scripts.

 - protG/anneal.inp: fixed bug in vdw loop.

 - PASD/Marvin: fixups in eginput/marvin/il4 example data.

 - improvements to runAll script.

Low-level changes:

 - RandomNum: added integerInRange() member.
 
 - jCoupPot:
   fixed variance calculation.
 
 - AtomSel: added reevaluate, constainsAtom and isMonomericallyEqualTo
   members.
 
 - Pot: added resetPotName() method.
 
 - IVM:
   small changes to Powell minimizer.
   added vel/setVel accessors.
 
 - cdsVector module:
   added sqrt function.
 
 - VMD:
   better cleanup of vmd atom selections.
 
 - EnsembleSimulation:
     added input multiplexer for the case in which stdin is a tty. This
     allows proper interactive behavior for numthreads>1, and should help
     with debugging.



----------------------------------------------------------------------
		     Changes from 2.9.9 to 2.10

general:

  new parameter/topology file naming convention:
    NMR protein refinement should now use protein.top / protein.par
  new command: tclXplor
    calls xplor -tcl. Can be used as command interpreter
  Darwin/ MacOSX change: version number now only includes the first digit. 
    This so the arch/equivList doesn't need to be updated so often.
  Darwin: 
    fixed Xplor-NIH-as-tcl-extension functionality
  xplor master script: 
    - now can specify parallel machines using the MACHINES env variable
      instead of the -machines option on the command line.
    - updated Linux ARCH def to deal with minor dot release numbers.
    - IRIX: properly treat LD_LIBRARYN32_PATH if it is set.
  eginputs:
   - added runAll script to run (most) scripts in eginputs.
   - fixed up scripts to use protein.top/par files.
   - minor updates to protG/anneal.inp

  
  
Python Interface:

  EnsembleSimulation:
    - can now specify nonuniform weighting.
    - fixed issue in which coordinate changes made in XPLOR shell were
      not propagated back to EnsembleSimulation.
  
  PotList:
    - change in behavior when an attempt is made to add a term with an
      instanceName already contained in the list: now the code throws an
      exception (previously, the second term was silently ignored).

  OrderPot:
   - new potential term to enable refinement again order parameters.
  
  CSAPot:
   - atomOrder accessor changed to support 15N CSAs.
   - fixed variance calc.
  
  AtomProb:
   - now normalize output map so that the largest map value is 1.
  
  PrePot:
   - update from Junji Iwahara
  
  ShapePot:
   - fixes- deal better with apparent singularities in angle diff. calc.
   - new degenerateTol member- determination of eigenvector/value
     pairing is now modified close to degeneracy- chose such that
     eigenvectors are those which align most closely with those in
     ensemble member 0.
   - added Pairwise target type- minimize pairwise differences.
  
  Python CDSVector:
    added fromList accessor.
  
  ivm:
    - added stepsizeThreshold member. If set, will try to avoid small
      timesteps by performing minimization when the stepsize falls below
      this value.
  
  mat3:
   - added trace function.
   - added rows()/cols() accessors.
  
  protocol: 
    initParams(): 
      - added optional weak_omega argument to allow
        small deformations of the peptide bond.
      - added optional reset parameter to clear any previous parameters
    fixupCovalentGeom():
      - made a bit more robust.
    genExtendedStructure():
      - added pdbFilename argument
      - fixed up the use of the optional sel argument.
    added covalentMinimize(), initHBDA() helpers.

  psfGen:
    pdbToPSF(): the pdbRecord argument may now be a filename instead of
                a string
    seqToPSF(): added optional deprotonateHIS argument.
    added cisPeptide() function to make cis-peptide bonds.
  
  ShapePot:
    added analyze tools.
  
  SimulationTools:
    better handling of analysis of existing files. No need to specify 
       numStructs. 
    added testGradient() helper.
  
  JCoupPot:
    added jCoupPotTools including create_JCoupPot() and analyze() helpers.
  
  PotList:
    analyze summary now prints sum of violations, not average.
  
  varTensorTools:
    added massSetup() helper.
    analyze(): print out more Da/Rh info, if using an ensemble.

XPLOR interface:

  hbdb update from Alex Grishaev:  
   adds ability use multiple segments
   other fixes to prevent crashes
  
  fixup time reporting under AMD64, Solaris

  SET MESSages statement: 
    now allow NONE to be a valid keyword (same as OFF).
 
  xplor command: allow XPLOR script to be specified as an argument. It
                 no longer needs to be redirected through stdin.

----------------------------------------------------------------------
		     Changes from 2.9.8 to 2.9.9

XPLOR interface:

  fixed problem in which Xplor-NIH would clobber /dev/null if the program was 
     run as root.
  plane.fcm: removed static arrays from planarity restraints.

general:
  configure script: added -symlinks option
  pdf2psf: fixup in reading arguments.
  seq2psf: added disulfide bond support with 
           -disulfide_bond and -disulfide_bridge options

PASD/Marvin fixup for problems at startup.

Python Interface

  updated eginputs/protG/anneal.py
  cdsVector, mat3, vec3 updates. Need documentation.
  python interpreter speedups (large) when tracing is off.
  python: added version, version_info to xplor module.
  XplorPot term now supports the scale() accessor.

----------------------------------------------------------------------
		     Changes from 2.9.7 to 2.9.8


XPLOR interface

  added support for set PRINT=off end statement, to suppress (most) output.
  related- now reroute much output through PUNIT instead of unit 6 (stdout).


Python interface

  protocol.py:
    added 
     genExtendedStructure() - generates a random initial structure.
     fixupCovalentGeom()    - corrects bonds, angle, dihe, atom overlap.
    modified
     initNBond()       - added optional nbxmod, tolerance parameters.
     torsionTopology() - added optional parameter to fix omega angle.

  psfGen.py:
    re-enabled and fixed the terminal atom chain break code
    seqToPSF(): now takes sequence as list or string
  
  rdcPotTools:
    added support for CAHN expts. provided by Alex Maltsev
  
  ringGeom, structureTools: structure analysis tools.
  
  selectTools:
    IVM_groupRigidSidechain(), convertToSelectionIndices():
      optimizations
    updated the following to correctly deal with multiple segments:
      IVM_breakProlines, IVM_breakRiboses, IVM_breakDisulfides, 
      IVM_groupRigidBackbone, groupRigidSel
    rigidProteinSelections: 
      ARG entry:   removed CE atom from rigid group list
  
  simulationTools:
    StructureLoop constructor: added the startStructure argument, fixes
  
  trace:
    fixes to work better with non-XplorNIH python interpreter
  
  varTensorTools:
    calcTensor(): fixes, now works with CSA expts too.

  replaced readline with editline (license issues).

  pyXplor script now treats cmd line arguments correctly.

  PDBTool: fixed a memory leak, added addRemarks() method to Python interface.


PASD/Marvin auto-assignment

  added AENEAS (Automated Experimental NoE Assignments based on Shifts):
    -accepts chemical shift, NOE peak position tables, matches them up, 
    filters them in reasonable ways, to produce an initial NOE restraint table.
    -now the primary input mechanism for Marvin.  
    -examples of use in the interleukin-4 and cyanovirin example cases.

  Support for 3dC, 3dN, 4dCC and 4dCN NOESY data

  Automatic detection of solvent lines

  Terminology change:  Rather than refer to, for example, Hi, Ci, 
    and Hj in a 3dC experiment, they're now referred to as fromProton, 
    fromCarbon, and toProton.  

  Reduced output during covalent geometry cleanup.

  Improved energy minimization.


Other various potential terms:

  Ensemble calculations:
    children now ignore the terminal suspend signal.
    performance, robustness improvements.
  
  CSAPot:
    fixups, cleanups
    restraint: corrected the tensor() method
  
  NOEPot:
    fixed segfault when reading a bad restraint (in code compiled with
    Intel compilers).
  
  RDCWavePot: 
    bugfixes
  
  ShapePot: 
    performance improvements.

  paraRestraints code fixes


Examples/Documentation

  documentation additions, fixups
  protein G example:
    update anneal.py: cleanup, modernization. Now psf and initial coords are
    generated on the fly from sequence info.
    added tang.py script for pretty printing of torsion angles from an
    ensemble of structures.

Architecture specific notes:

  combined Darwin+other os distribution should now work.
  Darwin build: 
    python libs now linked against xplor executable, rather than shared
    libs so that extension module has a better chance of working.

  Linux builds:
    workaround for segfaults in recent Redhat 2.4 kernels which incorporate 
    the back-ported thread library.

    The 2.2 kernel binary package has been discontinued. Please
    contact us if you are affected by this decision.

Parameters

  parnah1er1_mod_new.inp:  contribution from Zeena Johar:
    new impropers added to define chirality of ribose carbons


Running Xplor-NIH

  added xplor_mpi - helper script for running Xplor-NIH under MPI.
  testDist: now doesn't clean up if a failure occurs.

Low-level code

  CDSMap- added iterator over key/value pairs

  CDSStringStream, fdstream- 
    fix for memory leak (beyond my control) under Intel compilers

  FixedMatrix- added setDiag(vector)

  added EnsemblePot base class for ensemble potentials
    allows for code reuse and simplifies the use of synchronization
    barriers, with better performance characteristics.

  Modified update: support for telling which aspect is modified
  (simulation or potential variable)

  XplorWrap: add pid to temp filenames (used to communicate with XPLOR),
    to avoid resource contention. (e.g. MPI jobs on local host).

  support for SWIG version 1.3.22


     ------------------------------------------------------------
		     Changes from 2.9.6 to 2.9.7

General changes

  new documentation page at http://nmr.cit.nih.gov/xplor-nih/doc/current/

  documentation page with update Python documentation

  also available at the command-line via xplor -pydoc

  XPLOR manual updates
	
  added two scripts: 
    -bin/pdf2psf to generate a structure file from a pdb file.
    -bin/pyXplor to facilitate using Xplor-NIH as a Python interpreter.

  new architecture supported: AMD64/X86_64 (Opteron/Athlon64).


Python interface changes:

  protocol.py: updated, added torsionTopology()

  selectTools.py: added getResids()

  surfaceTools.py: added calcSA()

  rdc/csa orientational tensor setup for IVM simplified with freedom()
          accessor, and varTensorTools.topologySetup()


XPLOR interface changes

  define four variables: $echo $prev_echo, $messages $prev_messages
    to better control echoing, messaging.
  
  suppress many messages when messages==off

  csa.f: fixed rmsd calculation

  segmnt.f: added NUMBer statement to the SEGMent statement - to allow
            starting resid to be specified - without renumbering.


Marvin/PASD changes: 
  
  allow import from nmrDraw, NMRSTAR, XEASY, and PIPP data formats.

  initial matchup of NOE peak locations and chemical shift tables is 
    now handled by Marvin:
    - STAPP no longer needed.
    - symmetry-related peaks are detected.
    - restraints/assignments fully annotated with NOE peak, chemical shift
      source info.

  eginputs/marvin/cvn and eginputs/marvin/il4 were re-worked
    to illustrate the new data import facility.


Parameter changes

  topallhdg.dna: swapped ADE H61|H62 positions and CYT H41|H42 positions


Other changes

  ensemble calculations: -improved performance by not blocking instead of
                          spin for num_threads>1
                         -added singleThread(), multiThread() methods
                         -better treatment of irregular shutdowns
                         -subtle IVM bug: different ensemble members could
                          have different temperatures.
                         -fixup for JCoupPot for Ne>1
                         -new ability to share python objects across an
			  ensemble.

  surfaceArea: many fixes and improvements including approximations and 
               data-caching to improve performance.

  pdbTool: xplorAtomName now better translates * characters.
           will now read/write to a string buffer.

  added rms() and violations() as required entries in the Pot interface




     ------------------------------------------------------------
		     Changes from 2.9.5 to 2.9.6

General changes

 removed limits on all PSF parameters- the arrays have been moved from
 Fortran common blocks to the heap. Xplor-NIH startup size is now about
 25MB smaller, and it will handle systems as large as your computer's
 memory will allow. 

 Removed the CSTACK- and the associated summary line at exit.

PASD/Marvin
  - improved memory cleanup
  - improvements to peak picking code.
  - general fixups.

Developer changes

  added source/i1mach.c for machine constants

  XplorVars changed to a singleton class

  removal of SWIG type registration code- replaced by calls to TypeQuery.

     ------------------------------------------------------------
		     Changes from 2.9.4 to 2.9.5

General changes

 merged isac code for floating RDC alignment tensor.
   [from H.J. Sass, G. Musco, S.J. Stahl, P.T. Wingfield and S.Grzesiek,
     J. Biomol. NMR 21: 275-280 (2001). ]

 merged hbdb code from A. Grishaev. 
   ``An empirical pseudo-potential that encodes for the relative
     arrangement of two protein peptidyl units linked by a
     backbone-backbone hydrogen bond.''

  faster program startup by lazy initialization - only start
  python/tcl when they're needed.

  bin/testDist now prints and tallies CPU times of the tests

  Linux_2.4_i686 build verified working on 2.6 kernels, and on AMD64 (opteron) 
  processors.

Python interface:

  added CSA potential.

  reworked RDC potential (old scripts will not work - please see
                          eginput/protG/anneal.py for examples of how to
                          fix)
    [tensor moved to a separate VarTensor class ]

  added Gauss-Bonet surface area (with gradient) calculator. See 
  python/tests/saTest.py

  protocol.py:
    added initParams(), initStruct().
    updated initMinimize, initDynamics to be more complete.

  selectTools: 
    added minResid()

  simulationTools:
    added pdbFile() method which generates a pdb file name from a template.

  pdbTool:
    updated so that the filename() accessor actually works.

  XplorPot() - can now specify argument as lower case.

  general:
     atom selection arguments can now be specified as a python string

TCL interface:

  PASD/marvin automatic NOE assignment package updates.

  added DerivList wrapper

Parameters:

  updated dna/rna parameters.

Developer Changes:

 CDSlib update:
  Complex type updates
  cdsList: added pop() method, can now interact with generic vector
  cdsVector: can now interact with generic vector
  cdsSStream: added str_volatile() (old behavior), changed str() to return
              a copy of the streambuf
  chull: convex hull calculator class
  matrixTools: added setRow/Column manipulators

 source ( Fortran source ) updates:
  README: includes style rules for XPLOR Fortran submissions.
  freehp: now clear the heap index (set it to zero).
  added ability to turn off vdw energy

     ------------------------------------------------------------
		     Changes from 2.9.3 to 2.9.4

General changes

  merged the PARArestraints module from  
    http://www.postgenomicnmr.net/SW/PARArestraints.html

  new script: bin/seq2psf to generate psf file from sequence
              do seq2psf --help for info.

  Fixup so compilation with NOTCL=1 actually works

  proton chemical shifts: added setup file to databases/protshifts,
     example in test/protoncshift2.inp

Examples and Tests

  eginput: 
     updated: mef_dna, analysis_SRY, dock_dipolar_chemshift, sry,
	    marvin, DNA-SRY, PSF_generation, SRY_data_preprocessing

     marvin: added cvn example, other script fixups

     protG: important fixes to anneal.inp, anneal.py
  
  Unit tests:
    test: updated carboncshift.inp,
          added    protonchift2.inp, paraRestr.inp, pyCmd.inp, tclCmd.inp
    python: added matTest.py, protocolTest.py
            updated rdcPotTest.py

Python Interface changes:

  important bug fix for IVM

  improved analysis capabilities. See e.g. eginput/protG/anneal.py

  update to RDC potential. Added calcTensor, makeTable helpers

  added additional linear algebra capability

  cleanup of distributed python directory

  now built with Python-2.3.3

  new Dihedral class for measuring dihedral angles

  updated help entries


TCL Interface changes:

  marvin: updates, fixes

  interface fixups

Platform specific:

  Darwin port: support for newer OS versions
               better support for IBM xlf

  ia64 port: support for version 8 of Intel compilers
             fix \0 characters in username

  x86 port: fixups for compilation with pgf77 and g77

Developer changes:

  improved distclean rule

  update to classes: CDSMatrix, auto_ptr, 

  new datastructures, algorithms:
    circular doubly-linked list, convex hull template algorithm
    symmetric heap matrices+support

  support added for swig-1.3.21, but it seems to have problems.

     ------------------------------------------------------------
		     Changes from 2.9.2 to 2.9.3

Core XPLOR code

  Darwin port: now compiled with the ibm xlf compiler- with a big
               speed improvement. Many improvements to build more
               easily with and without fink. A big thanks goes to 
               Jack Howarth (howarth -at- bromo.med.uc.edu) for doing
               most of this work. 

General changes

  eginput: updates, and addition of new example scripts.
  
  ia64 port is improved: full regression tests now report a single
                         error (in the XPLOR interface).


Python Interface changes

 - includes paramagnetic relaxation enhancement module based on the
   modified Solomon-Bloembergen equation and 
   multiple structure representation for paramagnetic groups. This module
   contributed by Junji Iwahara. Please contact him at 
   iwahara-at-helix.nih.gov(replace -at- with a @) for further information.  

 - RDCPotTools:
    -- added the following methods: saupeMatrix, normalizedScalarProduct

 - PotList interface improved to allow use of a simple Python sequence
   in most instances. Also, the potentials returned by indexing a
   PotList object are now cast back to their original type (instead of
   being generic Pots).

 - SimulationTools:
     added analyze() function to analyze terms of a potList and return
     a concise summary appropriate for the remarks section of a PDB file.

 - can now access argv.

 - help files updated.

TCL Interface changes

 - update to interactive interface.

 - can now access argv.

Developer changes

  initial entry point changed from Fortran to C (xplor.c)- so that
  argv is more easily used.

  upgraded to pcre-4.4

  added symlink support for Darwin-7.0, Darwin-7.2

  added XPLOR tests for the TCL and Python interfaces.

  improved Fortran rtl initialization, when needed.



     ------------------------------------------------------------
		     Changes from 2.9.1 to 2.9.2


Core XPLOR code

  constraint inter code: fixed bug encountered when a double selection contains 
                         more than 9999 atoms
  
  ia32/linux port: fixed bug in which \000 would sometimes be written in
                   the username output
  
  sani potential: fixed bug encountered when using the ``square well'' potential
                type

General changes

  a FAQ is included in helplib/faq
  
  eginput: updated protG .inp, .py scripts
           many changes and additions for other examples.
  
  ia64 port is better, but full regression tests still report two errors.


Python Interface changes

  rdcPot: now supports averaging of dipolar couplings
          rdcPotTools.py updated
  
  selectTools.py updated
  
  simulationTools.py: StructureLoop now will deal appropriately if the number
                      of structures is smaller than the number of processes
  
  vmdInter.py: added loadFiles() helper for loading ensemble of structures


Developer changes

  added new install rule. Define INSTALL_DIR in Makefile.loc and make install
    creates an installation of Xplor-NIH separate from that in the source
    directory.
  
  iostream .h headers are now wrapped by .hh versions in CDSlib
  
  symbols in cdsMath.hh changed namespace from CDS to CDSMath
  
  matrixTools: svd matrix size fixed in python interface
  
  moved tcl/python test scripts to tests subdirectories
  
  enumNameMap.hh: added functions to help with conversions
  
  reworked Simulation id code


VMD-XPLOR interface

  py-xplor: now explicitly set intr char to ^C. This had caused problems
            with vmd-xplor on SGIs

  vmdInter code: made color setting more robust.