Index of /cvmfsmonitor/repository/wenmr.egi.eu/BMRZ/mars/1.1.3/html

	Name	Last modified	Size	Description
	Parent Directory		-
	image/	2005-03-18 07:52	-

As described in the papers:

1. Mars - robust automatic backbone assignment of proteins
   
2. Backbone assignment of proteins with known structure using residual dipolar couplings
   
   Young-Sang Jung and Markus Zweckstetter, submitted.

What is MARS?

MARS is a program for robust automatic backbone assignment of ¹³C/¹⁵N labeled proteins. MARS simultaneously optimizes the local and global quality of assignment to minimize propagation of initial assignment errors and to extract reliable assignments. It can be applied independent of the assignment complexity, it does not require tight thresholds for establishing sequential connectivity or detailed adjustment of these thresholds, it can work with a wide variety of NMR experiments and it is robust against missing chemical shift information.








Key features

1. simultaneous optimization of the local and global quality of assignment
2. exhaustive search for fragment lengths comprising up to five PRs during linking and mapping
3. best-first elements for both linking and mapping
4. combination of the secondary structure prediction program PSIPRED (McGuffin et al., 2000) with statistical chemical shift distributions, which were corrected for neighboring residue effects (Wang and Jardetzky, 2002), to improve identification of likely positions in the primary sequence
5. assessment of the reliability of fragment mapping by performing multiple assignment runs with noise-disturbed chemical shifts

Download

MARS

MacOS X
SGI Irix 6.2 version
Linux




The download provides a compressed tar archive with a MARS and PALES executable and example files. The archive can be unpacked with a command like the following:

zcat MARS.linux.tar.Z | tar xvf -
zcat pales.linux.tar.Z | tar xvf -
		




Users are encouraged to email the author to be informed about updates and related software.

PALES (home)

For assignment using residual dipolar couplings the software PALES has to be installed in addition. This is only required if a 3D structure is known.

Setup

Open your .cshrc files in your home directory. Add the three lines below to your .cshrc file.



	setenv MARSHOME  directoryName
	setenv PALESHOME  directoryName
	alias runmars directoryName/runmars

(directoryName is the name of the directory that contains the binary and script files.)

An example:

	setenv PALESHOME  /usr/users/yjung/bin/PALES
	setenv MARSHOME   /usr/users/yjung/bin/MARS
	alias runmars    '/usr/users/yjung/bin/MARS/runmars'
	

Getting started

Input


MARS is a program for backbone assignment of 13C/15N labeled proteins. Accordingly, following input is required:


• Obligatory
  
  
  • parameter setup file (mars.inp)
  • chemical shift table (SPARKY format)
  • primary sequence (FASTA format)
  • secondary structure prediction file (PSIPRED format)
    
    
    
    When a 3D structure is known and RDCs values are available
    
    
  • PDB file
  • RDC table (PALES format)
    
    
    
• Optional
  
  
  • table that allows restriction of the amino acid type and/or fixing of an assignment
  • table that allows fixing of sequential connectivities between pseudoresidues.



How to run Mars

1. Prepare your chemical shift table.
2. Get your primary sequence in FASTA format.
3. Get a secondary structure prediction using the Psipred web server.
4. Adjust the parameter setup file (mars.inp).
5. Type 'runmars mars.inp'
   

Output

• Assignment result filtered for high, medium and low reliablity ('assignment_AA.out').
  
  
• Assignment result including alternative assignments that show up with a 10 % probability ('assignment_AAs.out').
• The most likly assignment for each pseudoresidue ('assignment_PR.out').
• Summary of all possible connectivities ('connectivity.out').
• Summary of reduced possible connectivities ('connectivity_reduced.out').
• Chemical shift table with updated assignments that can be read into SPARKY ('sparky_all.out').
  
  
• Detailed information about predicted chemical shifts, number of reliable assignments, number of constraints for each pseudoresidue, matrices matching experimental and back-calculated chemical shifts and/or RDCs and pseudoenergy matrices at each iteration step ('mars.log').

Setting up input files

Obligatory



1. A Mars run is controlled by the parameter setup file (mars.inp). This has to be adjusted to the available experimental data. Please see below for a detailed description of the parameters.
   
   
   Lines with a '#' sign as first character as well as empty line are ignored.
   Do not change the variable names such as nIter.

   	
   mars.inp (MARSHOME/example/noStructure/1ubq/input)
   				
   
   	
   fragSize: 	5   				# Maximum length of pseudoresidue fragments
   cutoffCO: 	0.25				# Connectivity cutoff (ppm) of CO [0.25]
   cutoffCA: 	0.2										# Connectivity cutoff (ppm) of CA [0.5]
   cutoffCB: 	0.5										# Connectivity cutoff (ppm) of CB [0.5]
   cutoffHA: 	0.25									# Connectivity cutoff (ppm) of HA [0.25]
   
   
   fixConn: 	fix_con.tab			# Table for fixing sequential connectivity
   fixAss: 	fix_ass.tab 			# Table for fixing residue type and(or) assignment
   
   
   pdb: 		0   				# 3D structure available [0/1] 
   resolution: 	NO					# Resolution of 3D structure [Angstrom]
   pdbName: 	NO					# Name of PDB file (protons required!)
   tensor: 	NO	   				# Method for obtaining alignment tensor [0/1/2/3/4]
   nIter: 		NO  				# Number of iterations [2/3/4]
   	
   dObsExh: 	NO					# Name of RDC table for exhaustive SVD (PALES format)
   dcTab: 		NO					# Name of RDC table (PALES format)
   
   deuterated: 	0 	  				# Protonated proteins [0]; perdeuterated proteins [1]
   sequence: 	1ubq_fasta.tab 		# Primary sequence (FASTA format)
   secondary: 	1ubq_psipred.tab	# Secondary structure (PSIPRED format)
   csTab: 		1ubq_cs.tab        	# Chemical shift table
   			
   		

2. chemical shift table
   
   
   The chemical shift table follows the SPARKY format. It consists of a header, pseudoresidues and chemical shifts. The header has to be defined before the listing of chemical shift values starts and includes the variable names for the chemical shifts. Currently 10 different chemical shifts are supported and should be indicated by 'CA', 'CA-1', 'CB', 'CB-1', 'CO', 'CO-1', 'HA', 'HA-1', 'H' and 'N'. These variable names have to be in the same order as the columns for the different chemical shifts. The first column has to be the pseudoresidue column and other columns are chemical shift columns. Pseudoresidue means the name of the group of peaks which share the same (or similar due to the experimental imperfection)  N and HN chemical shifts. Lines with a '#' sign as first character as well as empty lines are ignored. Missing chemical shift values have to be indicated by ' - '.
   
   

    
   1ubq_cs.tab(MARSHOME/example/noStructure/1ubq/1ubq_cs.tab)
   		
   
   
   			N		CO-1		H		CA-1		CA
   	PR_2 		123.220		170.540		8.900		54.450		55.080
   	PR_3? 		115.340		175.920		8.320		55.080		-
   	PR_4?		118.110		172.450		8.610		59.570		55.210
   	PR_5GLY		121.000		175.320		9.300		55.210		60.620
   	PR_6GLY		127.520		-      		8.820		60.620		54.520
   	PR_7		115.400		177.140		8.730		54.520		60.470
   	PR_8		121.330		176.910		9.100		60.470		57.580
   	PR_9		105.590		178.800		7.630		57.580		61.400
   	PR_10??		108.890		175.520		7.810		61.400		45.460
   	:
   	:
   	:
   
   

   Pseudoresidue names can consist of characters+number+characters and characters are optional. Pseudoresidues are only distingushed by the number, which can be any integer number larger than 0. Therefore following, equivalent notations are allowed for a pseudoresidue:
   
   
   Correct format but equivalent to each other:

   
   	PR_37, g_37, 037, PR_037, 37GLY, PR_37???, ...
   

   
   
   Correct format for pseudoresidues:

   
   	37, g01, g135, K016, LLR001,g12?, 124D, ...
   

   
   
   Incorrect format for pseudoresidues:

   
   	34PR_37, g01_1, PR_34_1, g135GLY1, ...
   

   
   
   
   
3. The primary sequence of the protein has to be in FASTA format.
   
   
   IMPORTANT: 'X' and 'Z' can not be used for the characters of a sequence.
   
   

   1ubq_fasta.tab (MARSHOME/example/noStructure/1ubq/1ubq_fasta.tab)
   				
   
   > ubq
   MQIFVKTLTGKTITLEVEPSDTIENVKAKIQDKEGIPPDQQRLIFAGKQLEDGRTLSDYN
   IQKESTLHLVLRLRGG
   				
   			

4. Secondary structure prediction table
   
   
   This has to be in Psipred format. Use the Psipred web server to get the table.
   

   1ubq_psipred.tab (MARSHOME/example/noStructure/1ubq/1ubq_psipred.tab)
   	
   
   PSIPRED PREDICTION RESULTS
   
   Key
   
   Conf: Confidence (0=low, 9=high)
   Pred: Predicted secondary structure (H=helix, E=strand, C=coil)
     AA: Target sequence
   
       
   
   Conf: 968896699888999867863189999999997689875658887777738887136726
   Pred: CEEEEECCCCCEEEEEECCCCCHHHHHHHHHHHHCCCHHHEEEEECCEECCCCCCHHHHC
     AA: MQIFVKTLTGKTITLEVEPSDTIENVKAKIQDKEGIPPDQQRLIFAGKQLEDGRTLSDYN
                 10        20        30        40        50        60
   
   
   Conf: 8988889999950699
   Pred: CCCCCEEEEEEECCCC
     AA: IQKESTLHLVLRLRGG
                 70
   			
                 
   		

   If a 3D structure and experimental RDCs are available:
   

5. PDB file
   
   
   All standard PDB files can be used (including MOLMOL files).
   
   IMPORTANT: When using shape-prediction all atoms in the PDB file will be used including pseudo atoms (ANI).
   
   
   
6. Dipolar coupling input
   
   Experimental dipolar couplings are supplied according to the PALES table format:
   
   
   • The protein sequence should be given as shown by one or more "DATA SEQUENCE" lines. Space characters in the sequence will be ignored.
   • The table must include columns for residue ID, three-character residue name and the atom name for both atoms that are involved in the dipolar coupling as well as the dipolar coupling itself, its error and a weighting factor. Segment ID and Chain ID are optional.
     IMPROTANT: The atom notation must match that of the PDB file.
   • The table must include a "VARS" line that labels the corresponding columns of the table.
   • The table must include a "FORMAT" line that defines the data type of the corresponding columns of the table.
   • Lines with a '#' sign as first character as well as empty lines are ignored.
   
   
   
   Example dipolar coupling table (excerpts):

   DATA SEQUENCE MQIFVKTLTG KTITLEVEPS DTIENVKAKI QDKEGIPPDQ QRLIFAGKQL
   DATA SEQUENCE EDGRTLSDYN IQKESTLHLV LRLRGG
   
   VARS   RESID_I RESNAME_I ATOMNAME_I RESID_J RESNAME_J ATOMNAME_J D      DD    W
   FORMAT %5d     %6s       %6s        %5d     %6s       %6s    %9.3f   %9.3f %.2f
   
       2    GLN      N      2    GLN     HN     -15.524     1.000 1.00
       3    ILE      N      3    ILE     HN      10.521     1.000 1.00
       4    PHE      N      4    PHE     HN       9.648     1.000 1.00
       5    VAL      N      5    VAL     HN       6.082     1.000 1.00
   
       1    MET      C      2    GLN     HN       3.993     0.333 3.00
       2    GLN      C      3    ILE     HN      -5.646     0.333 3.00
       3    ILE      C      4    PHE     HN       1.041     0.333 3.00
       4    PHE      C      5    VAL     HN       0.835     0.333 3.00
   
       1    MET      C      2    GLN      N       2.651     0.125 8.00
       2    GLN      C      3    ILE      N      -3.768     0.125 8.00
       3    ILE      C      4    PHE      N       1.463     0.125 8.00
       4    PHE      C      5    VAL      N      -1.726     0.125 8.00
   
       2    GLN      N      2    GLN     HN     -15.524     1.000 1.00
       3    ILE      N      3    ILE     HN      10.521     1.000 1.00
       4    PHE      N      4    PHE     HN       9.648     1.000 1.00
       5    VAL      N      5    VAL     HN       6.082     1.000 1.00
   
       1    MET     HA      1    MET     CA     -38.341     1.000 0.50
       2    GLN     HA      2    GLN     CA      11.662     1.000 0.50
       3    ILE     HA      3    ILE     CA      18.424     1.000 0.50
       4    PHE     HA      4    PHE     CA      26.733     1.000 0.50
   
   





Optional

1. When additional information such as specific amino acid type labeling or initial manual assignments are available assignment of pseudoresidues can be restricted to single or to a set of residues. The first column has to be a pseudoresidue name followed by residue numbers or amino acid types to which the assignment should be restricted. Assignments can be fixed one by one by specipying the corresponding residue numbers or restrict it to a whole residue fragment by specifying the starting and ending residue number (inclusive) connected by '-' (without a blank in between the start and end number!). At the same time, amino acid types can be fixed by specifying the corresponding one letter code. More than one amino acid type can be specified by concatenation of the corresponding one letter codes (i.e. attach additional one-letter codes without blank in between).

   fix_ass.tab (MARSHOME/example/noStructure/1ubq/fix_ass.tab)
   	
   
   PR_3 3
   PR_10 10-15 23 34
   PR_12 12 34-36
   PR_13 13
   PR_14 14 16 HKT
   PR_15 LFR 66-69 13-16 9 71
   PR_16 EVA
   
   
   	
   

2. Also sequential connectivities can be fixed. This is especially useful when assignment is done iteratively by Mars and manually. The first and second column are pseudoresidue names. The first column is the name of the pseudoresidue for which the intra-residual chemical shift can be connected to the inter-residual chemical shift of the pesudoresidue in the second column.

   fix_con.tab (MARSHOME/example/noStructure/1ubq/fix_con.tab)
   	
   
   PR_2	PR_3
   PR_3	PR_4
   PR_4	PR_5
   PR_11   PR_12
   PR_12   PR_13
   PR_13   PR_14
   PR_25   PR_26
   PR_26   PR_27
   
   
   		
   	

Setting up assignment parameters

fragSize:

Sequential connectivity is established by matching inter- and intra-residual chemical shifts. Fragments comprising up to fragSize pseudoresidues are searched for exhaustively. The maximum segment length fragSize is a compromise between the desired total execution time of a MARS assignment run and the ability to reliably place PR segments onto the protein sequence.

According to our tests a fragSize of 5 is large enough to get reliable assignments (pseudoresidue fragments with length five can in most cases be placed uniquely into the protein sequence when intra- and inter-residual C^a and C^b chemical shifts are available).

For smaller proteins or if more computing power is available larger fragment sizes (six or seven) can be employed. This is expected to be useful if, for example, no C^b chemical shift information is available.



cutoff

• cutoffCO is the tolerance value (ppm) for matching intra- and inter-residual chemical shifts of C'.
• cutoffCA is the tolerance value (ppm) for matching intra- and inter-residual chemical shifts of C^a.
• cutoffCB is the tolerance value (ppm) for matching intra- and inter-residual chemical shifts of C^b.
• cutoffHA is the tolerance value (ppm) for matching intra- and inter-residual chemical shifts of H^a.

Cutoff values should be determined according to the resolution of the spectra. If chemical shifts were obtained from standard HNCACB, CBCACONH and HNCO experiments reasonable values will be

Ex.)
	cutoffCO: 0.1
	cutoffCA: 0.5
	cutoffCB: 0.5
	cutoffHA: 0.1
	

Note that too small error bounds will lead to a small number of reliable assignments.



fixConn

This is optional. If you want to fix sequential connectivities, prepare a table like fix_con.tab and specify the table name, otherwise set the fixConn parameter to NO.

NOTE: At one iteration step MARS generates 60 assignment solutions and extracts reliable assginments from these solutions. After the first iteration step MARS automatically fixes reliable assignments and reliable sequential connectivities obatined from previous iteration steps without user intervention. The iteration is continued until the number of reliable assignments does not increase any more. Therefore, one can see fixed assignments and fixed sequential connectivities on the screen during a MARS run although the user didn't fix anything at the start of MARS.

Ex.)

		fixConn: NO
	
			or
		
		fixConn: fix_conn.tab
	




deuterated

If a protein is perdeuterated, set the deuterated parameter to 1. Otherwise put it to 0.

Ex.)

		deuterated: 0
		
			or
			
		deuterated: 1		
	




sequence

Specifiy the name of the file that contains the primary sequence of your protein in FASTA format.

Ex.)

		sequence: 1ubq_fasta.tab
	




secondary

Specifiy the name of the file that contains the secondary structure information of your protein in PsiPred format.

Ex.)

		secondary: 1ubq_psipred.tab
	




csTab

Specifiy the name of the file that contains the experimental chemical shifts (SPARKY format).

Ex.)

		csTab: 1ubq_cs.tab
	






If no 3D structure or RDCs are available,
put the additional paramters as below:

pdb: 		0   			
resolution: 	NO			
pdbName: 	NO			
tensor: 	NO	   		
nIter: 		NO  			

dObsExh: 	NO			
dcTab: 		NO			
	





If a 3D structure and experimental RDC are available,
following parameters have to be set up:

pdb

Put the pdb flag pdb to 1, in order to use RDCs and the known 3D structure (otherwise set it to 0 ).

Ex.)

		pdb: 1
	




resolution

Specify the resolution of your crystal structure. If you don't konw the resolution of the structure because it is a homology model, set the resolution to ~ 4.0. In this case it will be useful to perform multiple assignment runs with decreasing values for the resolution parameter (suggested range is 2.0 < resolution < 6.0). The optimum value corresponds to the assignment run where the maximum number of reliable assignments was obtained.

Ex.)

		resolution: 1.8
	




pdbName

Name of file containing the coordinates of the 3D structure. All standard PDB files (including Molmol) can be used. IMPORTANT: Protons have to be present.

Ex.)
		pdbName: 1ubq.pdb
	




tensor

Method for obtaining an initial estimate of the alignment tensor. Four different modes are available that can automatically be accessed by specifying 1, 2, 3 or 4. The standard mode is 3.


If 1 is selected, MARS will use a gridSearch for estimating the orientation of the alignment tensor.
If 2 is selected, MARS will use exhaustive back-calculation (exhSVD). (dObsExh parameter has to be setup!)
If 3 is selected, MARS will use singular value decomposition (SVD).
If 4 is selected, MARS will use shape-prediction (shapePred).



Ex.)

		tensor: 3
		
		
				
	

(It is recommended to use 1 or 3 for the tensor parameter. Modes 2 and 4 require additional knowledge or RDCs in nearly neutral alignment media.)

nIter

MARS refines the initial alignment tensor estimate (obtained by the tensor method specified above) several times using SVD based on the reliable assignments obtained in previous iteration steps. Here, the number of refinement steps of the alignment tensor, nIter, can be defined. According to our tests 2 refinement steps are enough.

Ex.)

		nIter: 2
	




dObsExh

For exhaustive back-calculation (tensor mode 2) an RDC table is required that contains RDCs of a specific amino acid type. If the tensor mode is 1, 3 or 4, put the dObsExh parameter to NO.

Ex.)

		dObsExh: NO
		
			or
			
		dObsExh: dObs_1ubq_GLY.tab

	

dcTab

Name of file that contains the experimental RDC values (in PALES format).

Ex.)

		dcTab: dObs_1ubq.tab

Output

1. The first column is the residue number of the protein; the second column is the pseudoresidue that the residue is assigned to. The third column indicates the degree of reliability of each assignment. Three levels of reliability are distinguished:
   
   H indicates high reliablity as defined in the MARS paper. M and L do not fulfill all the criteria required for H reliablity and the specific criteria employed are adjusted automatically according to the completeness of the input data. Please see below for the robustness of assignments labelled as M and L.

   
   assignment_AA.out
   		
   
   MET_1      
   GLN_2      PR_2 (M)  
   ILE_3      PR_3 (M)  
   PHE_4      PR_4 (H)  
   VAL_5      PR_5 (H)  
   LYS_6      
   THR_7      
   LEU_8      
   THR_9      PR_9 (M)  
   GLY_10     PR_10 (H)  
   LYS_11     PR_11 (H)  
   THR_12     PR_12 (H)  
   ILE_13     PR_13 (H)  
   THR_14     PR_14 (H)  
   LEU_15     PR_15 (H)  
   GLU_16     PR_16 (M)  
   VAL_17     
   GLU_18     
   PRO_19     
   SER_20     PR_20 (L)  
   ASP_21     
   THR_22     
   	:
   	:
   	:     
   		
   

2. The first column is the residue number of the protein. Additional colums list pseudoresidues that can be assigned to this resdiue. Numbers in parenthesis are assignment probablities. Only pseudoresidues with an assignmnent probablity of higher than 10% are shown. assignment_AA.out is a subset of the assignments here.

   
   assignment_AAs.out
   		
   
   MET_1      
   GLN_2      PR_2 (96)  
   ILE_3      PR_3 (100)  
   PHE_4      PR_4 (100)  
   VAL_5      PR_5 (100)  
   LYS_6      PR_6 (63)  PR_8 (30)  
   THR_7      PR_7 (76)  
   LEU_8      PR_8 (61)  
   THR_9      PR_9 (100)  
   GLY_10     PR_10 (100)  
   LYS_11     PR_11 (100)  
   THR_12     PR_12 (100)  
   ILE_13     PR_13 (100)  
   THR_14     PR_14 (100)  
   LEU_15     PR_15 (100)  
   GLU_16     PR_16 (100)  
   VAL_17     PR_17 (73)  
   GLU_18     
   PRO_19     
   SER_20     PR_20 (86)  
   ASP_21     PR_21 (65)  
   THR_22     PR_57 (33)  
   	:
   	:
   	:  
   		
   

3. assignment_PR.out lists the most likely assignment for each pseudoresidue present in the input chemical shift table. The first column is the pseudoresidue and the second is the residue (to which the pseudoresidue can be assigned to most likely). NOTE: 'The most likely assignment' does not mean reliable assignment and two pseudoresidues can also be assigned to one residue. The information present in assignment_PR.out is useful if a pseudoresidue is not assigned to any residue in assignment_AAs.out and one asks himself what it might be assigned to.

   
   assignment_PR.out
   		
   
   PR_2 	 GLN_2   
   PR_3 	 ILE_3   
   PR_4 	 PHE_4   
   PR_5 	 VAL_5   
   PR_6 	 LYS_6   
   PR_7 	 THR_7   
   PR_8 	 LEU_8   
   PR_9 	 THR_9   
   PR_10 	 GLY_10  
   PR_11 	 LYS_11  
   PR_12 	 THR_12  
   PR_13 	 ILE_13  
   PR_14 	 THR_14  
   PR_15 	 LEU_15  
   PR_16 	 GLU_16  
   PR_17 	 VAL_17  
   PR_18 	 GLU_18  
   PR_20 	 GLN_40  
   PR_21 	 GLN_41  
   PR_22 	 SER_57
   	:
   	:
   	:
   	  
   		
   

4. In connectivity.out all possible sequential connectivities between pseudoresdiues are listed. All numbers are pseudoresidue numbers. The first column (closed by '-->') is the pseudoresidue number for which connectivities are listed. If no additional entries are present no connectivities could be found for that pseudoresidue. Otherwise, all pseudoresidue numbers are listed for which the inter-residual chemical shift can be matched to the intra-residual chemical shift of the pseudoresidue in the first column.

   
   connectivity.out
   		
   
   PR_2   -->	PR_3    PR_5    PR_35   PR_43   PR_69   PR_74   
   PR_3   -->	PR_4    PR_23   PR_30   PR_56   
   PR_4   -->	PR_3    PR_5    PR_29   PR_35   PR_43   
   PR_5   -->	PR_6    PR_8    PR_71   
   PR_6   -->	PR_2    PR_7    PR_49   PR_55   
   PR_7   -->	PR_6    PR_8    
   PR_8   -->	PR_9    PR_21   
   PR_9   -->	PR_10   
   PR_10  -->	PR_11   PR_48   PR_76   
   PR_11  -->	PR_12   PR_42   PR_75   
   PR_12  -->	PR_13   PR_24   PR_62   PR_67   
   PR_13  -->	PR_14   PR_32   
   PR_14  -->	PR_15   
   PR_15  -->	PR_16   PR_44   
   PR_16  -->	PR_17   PR_69   PR_74   
   PR_17  -->	PR_18   PR_60   
   PR_18  -->	PR_16   PR_47   
   PR_20  -->	PR_9    PR_21   
   PR_21  -->	PR_22   PR_40   PR_41   PR_50   PR_73   
   PR_22  -->	PR_4    PR_23   PR_30   PR_56   
   		
   		
   

5. In connectivity_reduced.out all possible sequential connectivities between pseudoresdiues are filtered for reliable assignments (i.e. it is a subset of connectivity.out).

   
   connectivity_reduced.out
   		
   
   PR_2   -->	PR_3    
   PR_3   -->	PR_4    
   PR_4   -->	PR_5    
   PR_5   -->	PR_6    PR_8    PR_71   
   PR_6   -->	PR_2    PR_7    PR_49   PR_55   
   PR_7   -->	PR_6    PR_8    
   PR_8   -->	PR_9    PR_21   
   PR_9   -->	PR_10   
   PR_10  -->	PR_11   
   PR_11  -->	PR_12   
   PR_12  -->	PR_13   
   PR_13  -->	PR_14   
   PR_14  -->	PR_15   
   PR_15  -->	PR_16   PR_44   
   PR_16  -->	PR_17   PR_69   PR_74   
   PR_17  -->	PR_18   PR_60   
   PR_18  -->	PR_16   
   PR_20  -->	PR_9    PR_21   
   PR_21  -->	PR_22   PR_40   PR_41   PR_50   PR_73   
   PR_22  -->	PR_23   PR_56   
   		
   		
   

6. Detailed information about assignment parameters, percentage of expected intra- and inter-residual Ca, Cb, Co and Ha chemical shifts present in the input chemical shift table, predicted chemical shifts, number of reliable assignments, number of constraints for each pseudoresidue, matrices matching experimental and back-calculated chemical shifts and/or RDCs and pseudoenergy matrices at each iteration step.

   
   mars.log
    
   
   ------------------------------------------------------------------------------------------
   fragSize: 	5   			# Maximum length of pseudoresidue fragments
   cutoffCO: 	0.25			# Connectivity cutoff (ppm) of CO [0.25]
   cutoffCA: 	0.2			# Connectivity cutoff (ppm) of CA [0.5]
   cutoffCB: 	0.5			# Connectivity cutoff (ppm) of CB [0.5]
   cutoffHA: 	0.25			# Connectivity cutoff (ppm) of HA [0.25]
   
   
   fixConn: 	fix_con.tab		# Table for fixing sequential connectivity
   fixAss: 	fix_ass.tab 		# Table for fixing residue type and(or) assignment
   
   
   pdb: 		0   			# 3D structure available [0/1] 
   resolution: 	NO			# Resolution of 3D structure [Angstrom]
   pdbName: 	NO			# Name of PDB file (protons required!)
   tensor: 	NO	   		# Method for obtaining alignment tensor [0/1/2/3/4]
   nIter: 		NO  			# Number of iterations [2/3/4]
   
   dObsExh: 	NO			# Name of RDC table for exhaustive SVD (PALES format)
   dcTab: 		NO			# Name of RDC table (PALES format)
   
   deuterated: 	0 	  		# Protonated proteins [0]; perdeuterated proteins [1]
   sequence: 	1ubq_fasta.tab 		# Primary sequence (FASTA format)
   secondary: 	1ubq_psipred.tab	# Secondary structure (PSIPRED format)
   csTab: 		1ubq_cs.tab        	# Chemical shift table
   ------------------------------------------------------------------------------------------
   
   # of AA: 76
   # of PRO: 3
   # of GLY: 6
   # of Assignable AA: 72
   # of PR: 72
   
   CA: 100.0   CB:   0.0   CO:   0.0   HA:   0.0
   Ca: 100.0   Cb:   0.0   Co: 100.0   Ha:   0.0
   
   
      AC    RC     RW
   --------------------
      53    11     0
   --------------------
      55    13     0
      54    13     0
   --------------------
      54    20     0
      53    20     0
   --------------------
   			:
   			:	
   			:
   			:
   			:
   			
   		
   

Analysis of assignment

1. Check the connectivity.out to verify that your chemical shift table has been made properly. If there are problems in your chemical shfit table due to miscalibration of spectra and(or) many pseudoresidues grouped incorrectly, you will see many missing sequential connectivities in the connectivity.out table.
   
   
2. MARS keeps reliablility of assignment even for highly degenerate and/or incomplete data sets and labels assignments according to three different levels of reliability, H (high reliable assignment), M (medium reliable assignment), L (low reliable assignment). The following table shows the robustness of Mars assignment for different proteins, different connectivity cutoffs and ranked according to H, M and L reliability. From this table it becomes clear that even medium reliable assignments are basically error-free and only for less stringent connectivity cutoffs assignments labelled as L contain a few errors.
   
   
   
   
   Table 1. (BMRB chemical shifts used for the test. / COcutoff: 0.25, CAcutoff: 0.5, CBcutoff: 0.5)

   Tested proteins	Used chemical shfits	# of AA (assignable AA)	H (Correct/Incorrect)	M (Correct/Incorrect)	L (Correct/Incorrect)
   Maltose binding protein	C', Ca, Cb	723 (654)	639/ 0	9/ 0	0/ 0
   Maltose binding protein	Ca, Cb	370 (335)	303/ 0	18/ 1	0/ 0
   EIN of the phospoenolpyruvate	Ca, Cb	259 (248)	232/ 0	6/ 0	0/ 0
   E-cardherin domains II	Ca, Cb	227 (167)	76/ 1	5/ 1	12/ 9
   Human prion protein	Ca, Cb	210 (190)	130/ 0	9/ 0	5/ 0
   Superoxide dismutase	C', Ca, Cb	192 (117)	101/ 0	2/ 1	2/ 0
   Calmodulin/M13 complex	C', Ca	148 (144)	37/ 0	7/ 0	17/ 2
   E. coli EmrE	C', Ca, Cb	110 (74)	35/ 0	6/ 0	10/ 3
   Human ubiquitin	Ca, Cb	76 (72)	72/ 0	0/ 0	0/ 0

   
   
   Table 2. (BMRB chemical shifts used for the test. / COcutoff: 0.15, CAcutoff: 0.2, CBcutoff: 0.4)

   Tested proteins	Used chemical shfits	# of AA (assignable AA)	H (Correct/Incorrect)	M (Correct/Incorrect)	L (Correct/Incorrect)
   Maltose binding protein	C', Ca, Cb	723 (654)	639/ 0	9/ 0	0/ 0
   Maltose binding protein	Ca, Cb	370 (335)	324/ 0	7/ 0	2/ 0
   EIN of the phospoenolpyruvate	Ca, Cb	259 (248)	246/ 0	0/ 0	0/ 0
   E-cardherin domains II	Ca, Cb	227 (167)	102/ 0	7/ 0	19/ 0
   Human prion protein	Ca, Cb	210 (190)	127/ 0	4/ 0	5/ 0
   Superoxide dismutase	C', Ca, Cb	192 (117)	104/ 0	0/ 0	1/ 0
   Calmodulin/M13 complex	C', Ca	148 (144)	142/ 0	0/ 0	0/ 0
   E. coli EmrE	C', Ca, Cb	110 (74)	58/ 0	4/ 0	0/ 0
   Human ubiquitin	Ca, Cb	76 (72)	72/ 0	0/ 0	0/ 0

   
   
   Table 3. (Raw peak lists used for the test. / CAcutoff: 0.3, CBcutoff: 0.5, HAcutoff: 0.05)

   Tested proteins	Used chemical shfits	# of AA (assignable AA)	H (Correct/Incorrect)	M (Correct/Incorrect)	L (Correct/Incorrect)
   Z domain protein	Ca, Cb, Ha	71 (67)	65/ 0	0/ 0	0/ 0
   Z domain protein	Ca, Cb	71 (67)	34/ 0	6/ 0	2/ 0

   
   
   
   
   
   
3. Connectivity_reduced.out filters all detected connectivities reported in connectivity.out according to high reliable assignments. This can be useful for manual refinement of a Mars assignment.
   
   
   

Directory structure

MARSHOME:

	
	$MARSHOME/bin
	$MARSHOME/example

	
	$MARSHOME/example/noStructure
	$MARSHOME/example/Structure


	$MARSHOME/example/noStructure/1ubq
	$MARSHOME/example/noStructure/1zym
	$MARSHOME/example/noStructure/1dmb

	
	$MARSHOME/example/Structure/1ubq
	$MARSHOME/example/Structure/1zym
	$MARSHOME/example/Structure/1dmb

Important points to remember

1. Spectra calibration and proper peak grouping are the most important points.
   
   
   
2. When grouping inter- and intra-chemical shifts try to use the same spectrum for extraction of inter- and intra-chemical shfits of a given atom type.
   
   For example, when you want to get intra- and interresidual chemical shifts of C^a, extract both chemical shifts from the HNCA spectrum. Only take the interresidual C^a chemical shift from a one way connectivity spectrum like HN(Co)CA, if the interresidual peak in the HNCA is too weak or overlapping.
   
   In that case, bigger connectivity cutoffs (cutoffCO, cutoffCA, cutoffCB, and cutoffHA) have to be used due to imperfections in spectrum calibration.
   
   Nevertheless, Mars does not care where you got the intra- and inter-chemical shifts from!
   
   
   
3. Be careful of folded peaks!
   
   

For advanced users

Tensor

Grid search


1116 alignment tensor orientations are systematically sampled and for each orientation the deviation D(i,j) between experimental and back-calculated RDCs is determined (equation 1).These 1116 orientations are obtained in a two-step procedure. First, the z-axis of the molecule samples uniformly 122 points on a unit sphere that were determined by a double cubic lattice method (Eisenhaber et al., 1995). Only one quarter of these points have to be taken into account due to the inversion symmetry of RDCs. In a second step, the molecule is rotated around the z-axis in steps of 10? giving a highly uniform and efficient sampling of all possible tensor orientations. All sampled orientations are ranked according to their corresponding D(i,j) values (equation 1) and the lowest D(i,j) value indicates the best estimate for the experimental alignment tensor. Thus, both chemical shifts and RDCs are used, making the extraction of the tensor orientation more robust.



Exhaustive SVD


In this method resonances belonging to a specific amino acid type are identified and all possible assignments are searched exhaustively for best-fit of back-calculated to experimental RDCs. The permutation that shows the best agreement identifies the experimental alignment tensor. exhSVD can be applied as long as at least one amino acid type is less than eleven times present in the primary sequence of the protein.



SVD


When the assignment is known, an alignment tensor can be obtained by best-fitting experimental dipolar couplings to a 3D structure using singular value decomposition (SVD) (Losonczi et al., 1999). Thus, a two-stage strategy for structure-enhanced assignment can be devised. In the first stage, RDCs are not used and partial assignments are obtained using only chemical shift matching or a combination of chemical shift matching and sequential connectivity information, i.e. using a common backbone assignment strategy. This assignment is assumed to be correct and a alignment tensor is obtained from best-fitting experimental RDCs to the 3D structure. The accuracy of the alignment tensor will depend on the percentage of correct assignments that were obtained in the first assignment phase. Tests show that, even when the percentage of correct assignment is below 50%, the alignment tensor is very close to its correct orientation. This is due to the fact that wrong assignments are not randomly distributed on the primary sequence of the protein. As Ca/Cb chemical shifts depend very much on the type of secondary structure, exchange of assignments mainly takes place between residues located on the same type of regular secondary structure. In addition, b-strands are often part of sheet structures, i.e. different strands are quite collinear. Therefore, residues located in these strands have similar RDCs and back-calculated alignment tensors are not severely affected by an assignment where pseudoresidues are interchanged between two b-strands (data not shown). The applicability of this approach depends mainly on the amount and type of chemical shift information available, i.e. how good is the assignment without RDCs.



Shape prediction


Second, molecular alignment tensors of proteins dissolved in nearly neutral bicelles can be predicted from the three-dimensional shape of the protein (Zweckstetter and Bax, 2000). This is particularly attractive as the only information necessary is the 3D structure, i.e. RDCs can be used in the same way as chemical shifts for structure-enhanced assignment. A disadvantage is that this method requires measurement of dipolar couplings in a nearly neutral alignment medium restricting the general applicability of RDC-enhanced assignment. However, recent results indicate that prediction of alignment tensors from 3D structures can be extended to a wide variety of alignment media, when electrostatic interactions are taken into account (Zweckstetter et al., 2003). Thus, this promises to be a valuable approach for obtaining approximate alignment tensors before the start of RDC-enhanced assignment.