SIGGEN documentation


 


CONTENTS

1.0 SUMMARY
2.0 INPUTS & OUTPUTS
3.0 INPUT FILE FORMAT
4.0 OUTPUT FILE FORMAT
5.0 DATA FILES
6.0 USAGE
7.0 KNOWN BUGS & WARNINGS
8.0 NOTES
9.0 DESCRIPTION
10.0 ALGORITHM
11.0 RELATED APPLICATIONS
12.0 DIAGNOSTIC ERROR MESSAGES
13.0 AUTHORS
14.0 REFERENCES



1.0 SUMMARY

Generates a sparse protein signature from an alignment


2.0 INPUTS & OUTPUTS

SIGGEN reads a directory of DAF files (domain alignment files) and, optionally, a directory of CON files (contacts file) containing a CON file for each aligned domain. It generates a sparse protein signature of a specified sparsity for each alignment. The base name of a signature file is the unique identifier (an integer) for the family, superfamily etc if one is specified in the DAF file, otherwise, the base name of the input DAF file is used. The paths of the input and output files are specified by the user and the file extensions are specified in the ACD file.


3.0 INPUT FILE FORMAT

The format of the domain alignment file is described in DOMAINALIGN documentation.


4.0 OUTPUT FILE FORMAT

The output file (Figure 1) uses the following records. Domain classification records for the node in SCOP or CATH from which the input alignment and therefore signature were derived are given. In this example, the four records taken from the DAF (input) file are CL, FO, SF and FA.

Two rows of data for the emprical residues and gaps are then given:

Output files for usage example

File: 54894.sig

TY   SCOP
XX
TS   1D
XX
CL   Alpha and beta proteins (a+b)
XX
FO   Ferredoxin-like
XX
SF   Aspartate carbamoyltransferase, Regulatory-chain, N-terminal domain
XX
FA   Aspartate carbamoyltransferase, Regulatory-chain, N-terminal domain
XX
SI   54894
XX
NP   15
XX
NN   [1]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   H ; 2
XX
GA   12 ; 2
XX
NN   [2]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   P ; 2
XX
GA   1 ; 2
XX
NN   [3]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   P ; 2
XX
GA   26 ; 2
XX
NN   [4]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   T ; 2
XX
GA   15 ; 2
XX
NN   [5]
XX


  [Part of this file has been deleted for brevity]

XX
GA   4 ; 2
XX
NN   [10]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   I ; 2
XX
GA   2 ; 2
XX
NN   [11]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   D ; 2
XX
GA   0 ; 2
XX
NN   [12]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   N ; 2
XX
GA   0 ; 2
XX
NN   [13]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   V ; 2
XX
GA   3 ; 2
XX
NN   [14]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   R ; 2
XX
GA   3 ; 2
XX
NN   [15]
XX
IN   NRES 1 ; NGAP 1 ; WSIZ 0
XX
AA   L ; 2
XX
GA   2 ; 2
//

File: 55074.sig

TY   SCOP
XX
TS   1D
XX
CL   Alpha and beta proteins (a+b)
XX
FO   Ferredoxin-like
XX
SF   Adenylyl and guanylyl cyclase catalytic domain
XX
FA   Adenylyl and guanylyl cyclase catalytic domain
XX
SI   55074
XX
NP   38
XX
NN   [1]
XX
IN   NRES 2 ; NGAP 2 ; WSIZ 0
XX
AA   H ; 1
AA   E ; 1
XX
GA   10 ; 1
GA   11 ; 1
XX
NN   [2]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   D ; 1
AA   T ; 1
XX
GA   1 ; 2
XX
NN   [3]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   I ; 1
AA   T ; 1
XX
GA   3 ; 2
XX
NN   [4]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   F ; 1
AA   I ; 1


  [Part of this file has been deleted for brevity]

AA   N ; 1
XX
GA   4 ; 2
XX
NN   [34]
XX
IN   NRES 2 ; NGAP 2 ; WSIZ 0
XX
AA   K ; 1
AA   A ; 1
XX
GA   4 ; 1
GA   8 ; 1
XX
NN   [35]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   W ; 1
AA   A ; 1
XX
GA   0 ; 2
XX
NN   [36]
XX
IN   NRES 2 ; NGAP 2 ; WSIZ 0
XX
AA   A ; 1
AA   T ; 1
XX
GA   14 ; 1
GA   16 ; 1
XX
NN   [37]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   K ; 1
AA   X ; 1
XX
GA   2 ; 2
XX
NN   [38]
XX
IN   NRES 2 ; NGAP 1 ; WSIZ 0
XX
AA   G ; 1
AA   D ; 1
XX
GA   1 ; 2
//




5.0 DATA FILES

SIGGEN requires a residue substitution matrix.


6.0 USAGE

6.1 COMMAND LINE ARGUMENTS

Generates a sparse protein signature from an alignment
Version: EMBOSS:6.3.0

   Standard (Mandatory) qualifiers (* if not always prompted):
  [-algpath]           dirlist    [./] This option specifies the location of
                                  DAF files (domain alignment files) (input).
                                  A 'domain alignment file' contains a
                                  sequence alignment of domains belonging to
                                  the same SCOP or CATH family (or other node
                                  in the structural hierarchies). The file is
                                  in DAF format (CLUSTAL-like) and is
                                  annotated with domain family classification
                                  information. The files generated by using
                                  SCOPALIGN will contain a structure-based
                                  sequence alignment of domains of known
                                  structure only. Such alignments can be
                                  extended with sequence relatives (of unknown
                                  structure) by using SEQALIGN.
   -mode               menu       [1] This option specifies the mode of
                                  signature generation. There are 3 modes for
                                  signatures generatation: (1) Use positions
                                  specified in alignment file. The alignment
                                  file must contain a line beginning with the
                                  text 'Positions' for each line of the
                                  alignment. A '1' in the 'Positions' line
                                  indicates that the signature should include
                                  data from the corresponding alignment site.
                                  The signature will only include the
                                  positions that are marked with a '1'. (2)
                                  Use a scoring method. The alignment is
                                  scored (see 'Algorithm') and the signature
                                  of a specified sparsity is sampled from high
                                  scoring positions. (3): Generate a
                                  randomised signature. A signature of a
                                  specified sparsity is sampled at random from
                                  the alignment. (Values: 1 (Use positions
                                  specified in alignment file); 2 (Use a
                                  scoring method); 3 (Generate a randomised
                                  signature))
*  -conoption          menu       [5] This option specifies the
                                  structure-based scoring scheme. SIGGEN
                                  provides 2 structure-based scoring schemes
                                  (plus a combination method) that are used to
                                  score the input alignment. (Values: 1
                                  (Number); 2 (Conservation); 3 (Number and
                                  conservation); 4 (None (structural data
                                  available)); 5 (None (no structural data
                                  available)))
*  -conpath            directory  [./] This option specifies the location of
                                  CON files (contact files) (input). A
                                  'contact file' contains contact data for a
                                  protein or a domain from SCOP or CATH, in
                                  the CON format (EMBL-like). The contacts may
                                  be intra-chain residue-residue, inter-chain
                                  residue-residue or residue-ligand. The
                                  files are generated by using CONTACTS,
                                  INTERFACE and SITES.
*  -cpdbpath           directory  [./] This option specifies the location of
                                  domain CCF files (clean coordinate files)
                                  (input). A 'clean cordinate file' contains
                                  protein coordinate and derived data for a
                                  single PDB file ('protein clean coordinate
                                  file') or a single domain from SCOP or CATH
                                  ('domain clean coordinate file'), in CCF
                                  format (EMBL-like). The files, generated by
                                  using PDBPARSE (PDB files) or DOMAINER
                                  (domains), contain 'cleaned-up' data that is
                                  self-consistent and error-corrected.
                                  Records for residue solvent accessibility
                                  and secondary structure are added to the
                                  file by using PDBPLUS.
*  -seqoption          menu       [3] This option specifies the sequence-based
                                  scoring scheme. SIGGEN provides 2
                                  sequence-based scoring schemes that are used
                                  to score the input alignment. (Values: 1
                                  (Substitution matrix); 2 (Residue class); 3
                                  (None))
*  -datafile           matrixf    [EBLOSUM62] This option specifies the the
                                  substitution matrix. The substitution matrix
                                  is used by the sequence-based scoring
                                  schemes.
*  -sparsity           integer    [10] This option specifies the % sparsity of
                                  signature. The signature sparsity is a
                                  user-defined parameter that determines how
                                  many residues the final signature will
                                  contain, for example, if the average
                                  sequence length of the proteins in the
                                  alignment is 250 residues, then a signature
                                  of sparsity 10% (default value) will contain
                                  25 key residues or signature positions,
                                  that correspond to the top 25% highest
                                  scoring alignment positions. (Any integer
                                  value)
   -wsiz               integer    [0] This option specifies the window size.
                                  When a signature is aligned to a protein
                                  sequence, the permissible gaps between two
                                  signature positions is determined by the
                                  empirical gaps and the window size. The user
                                  is prompted for a window size that is used
                                  for every position in the signature. Likely
                                  this is not optimal. A future implementation
                                  will provide a range of methods for
                                  generating values of window size depending
                                  upon the alignment (window size is
                                  identified by the WSIZ record in the
                                  signature output file). (Any integer value)
*  -filtercon          toggle     [N] This option specifies whether to
                                  disregard positions forming few contacts
                                  only during the selection of signature
                                  positions.
*  -conthresh          integer    [10] This option specifies the threshold
                                  contact number. This controls the selection
                                  of key positions for the structure-based
                                  scoring scheme (number of contacts). (Any
                                  integer value)
*  -[no]filterpsim     boolean    [Y] This option specifies whether to
                                  disregard alignment sites that were not
                                  aligned satisfactorily (STAMP alignments
                                  only).
  [-sigoutdir]         outdir     [./] This option specifies the location of
                                  signature files (output). A 'signature file'
                                  contains a sparse sequence signature
                                  suitable for use with the SIGSCAN and SIGGEN
                                  programs. The files are generated by using
                                  SIGGEN & SIGGENLIG.

   Additional (Optional) qualifiers: (none)
   Advanced (Unprompted) qualifiers: (none)
   Associated qualifiers:

   "-algpath" associated qualifiers
   -extension1         string     Default file extension

   "-conpath" associated qualifiers
   -extension          string     Default file extension

   "-cpdbpath" associated qualifiers
   -extension          string     Default file extension

   "-sigoutdir" associated qualifiers
   -extension2         string     Default file extension

   General qualifiers:
   -auto               boolean    Turn off prompts
   -stdout             boolean    Write first file to standard output
   -filter             boolean    Read first file from standard input, write
                                  first file to standard output
   -options            boolean    Prompt for standard and additional values
   -debug              boolean    Write debug output to program.dbg
   -verbose            boolean    Report some/full command line options
   -help               boolean    Report command line options and exit. More
                                  information on associated and general
                                  qualifiers can be found with -help -verbose
   -warning            boolean    Report warnings
   -error              boolean    Report errors
   -fatal              boolean    Report fatal errors
   -die                boolean    Report dying program messages
   -version            boolean    Report version number and exit

Qualifier Type Description Allowed values Default
Standard (Mandatory) qualifiers
[-algpath]
(Parameter 1)
dirlist This option specifies the location of DAF files (domain alignment files) (input). A 'domain alignment file' contains a sequence alignment of domains belonging to the same SCOP or CATH family (or other node in the structural hierarchies). The file is in DAF format (CLUSTAL-like) and is annotated with domain family classification information. The files generated by using SCOPALIGN will contain a structure-based sequence alignment of domains of known structure only. Such alignments can be extended with sequence relatives (of unknown structure) by using SEQALIGN. Directory with files ./
-mode list This option specifies the mode of signature generation. There are 3 modes for signatures generatation: (1) Use positions specified in alignment file. The alignment file must contain a line beginning with the text 'Positions' for each line of the alignment. A '1' in the 'Positions' line indicates that the signature should include data from the corresponding alignment site. The signature will only include the positions that are marked with a '1'. (2) Use a scoring method. The alignment is scored (see 'Algorithm') and the signature of a specified sparsity is sampled from high scoring positions. (3): Generate a randomised signature. A signature of a specified sparsity is sampled at random from the alignment.
1 (Use positions specified in alignment file)
2 (Use a scoring method)
3 (Generate a randomised signature)
1
-conoption list This option specifies the structure-based scoring scheme. SIGGEN provides 2 structure-based scoring schemes (plus a combination method) that are used to score the input alignment.
1 (Number)
2 (Conservation)
3 (Number and conservation)
4 (None (structural data available))
5 (None (no structural data available))
5
-conpath directory This option specifies the location of CON files (contact files) (input). A 'contact file' contains contact data for a protein or a domain from SCOP or CATH, in the CON format (EMBL-like). The contacts may be intra-chain residue-residue, inter-chain residue-residue or residue-ligand. The files are generated by using CONTACTS, INTERFACE and SITES. Directory ./
-cpdbpath directory This option specifies the location of domain CCF files (clean coordinate files) (input). A 'clean cordinate file' contains protein coordinate and derived data for a single PDB file ('protein clean coordinate file') or a single domain from SCOP or CATH ('domain clean coordinate file'), in CCF format (EMBL-like). The files, generated by using PDBPARSE (PDB files) or DOMAINER (domains), contain 'cleaned-up' data that is self-consistent and error-corrected. Records for residue solvent accessibility and secondary structure are added to the file by using PDBPLUS. Directory ./
-seqoption list This option specifies the sequence-based scoring scheme. SIGGEN provides 2 sequence-based scoring schemes that are used to score the input alignment.
1 (Substitution matrix)
2 (Residue class)
3 (None)
3
-datafile matrixf This option specifies the the substitution matrix. The substitution matrix is used by the sequence-based scoring schemes. Comparison matrix file in EMBOSS data path EBLOSUM62
-sparsity integer This option specifies the % sparsity of signature. The signature sparsity is a user-defined parameter that determines how many residues the final signature will contain, for example, if the average sequence length of the proteins in the alignment is 250 residues, then a signature of sparsity 10% (default value) will contain 25 key residues or signature positions, that correspond to the top 25% highest scoring alignment positions. Any integer value 10
-wsiz integer This option specifies the window size. When a signature is aligned to a protein sequence, the permissible gaps between two signature positions is determined by the empirical gaps and the window size. The user is prompted for a window size that is used for every position in the signature. Likely this is not optimal. A future implementation will provide a range of methods for generating values of window size depending upon the alignment (window size is identified by the WSIZ record in the signature output file). Any integer value 0
-filtercon toggle This option specifies whether to disregard positions forming few contacts only during the selection of signature positions. Toggle value Yes/No No
-conthresh integer This option specifies the threshold contact number. This controls the selection of key positions for the structure-based scoring scheme (number of contacts). Any integer value 10
-[no]filterpsim boolean This option specifies whether to disregard alignment sites that were not aligned satisfactorily (STAMP alignments only). Boolean value Yes/No Yes
[-sigoutdir]
(Parameter 2)
outdir This option specifies the location of signature files (output). A 'signature file' contains a sparse sequence signature suitable for use with the SIGSCAN and SIGGEN programs. The files are generated by using SIGGEN & SIGGENLIG. Output directory ./
Additional (Optional) qualifiers
(none)
Advanced (Unprompted) qualifiers
(none)
Associated qualifiers
"-algpath" associated dirlist qualifiers
-extension1
-extension_algpath
string Default file extension Any string daf
"-conpath" associated directory qualifiers
-extension string Default file extension Any string con
"-cpdbpath" associated directory qualifiers
-extension string Default file extension Any string ccf
"-sigoutdir" associated outdir qualifiers
-extension2
-extension_sigoutdir
string Default file extension Any string sig
General qualifiers
-auto boolean Turn off prompts Boolean value Yes/No N
-stdout boolean Write first file to standard output Boolean value Yes/No N
-filter boolean Read first file from standard input, write first file to standard output Boolean value Yes/No N
-options boolean Prompt for standard and additional values Boolean value Yes/No N
-debug boolean Write debug output to program.dbg Boolean value Yes/No N
-verbose boolean Report some/full command line options Boolean value Yes/No Y
-help boolean Report command line options and exit. More information on associated and general qualifiers can be found with -help -verbose Boolean value Yes/No N
-warning boolean Report warnings Boolean value Yes/No Y
-error boolean Report errors Boolean value Yes/No Y
-fatal boolean Report fatal errors Boolean value Yes/No Y
-die boolean Report dying program messages Boolean value Yes/No Y
-version boolean Report version number and exit Boolean value Yes/No N

6.2 EXAMPLE SESSION

An example of interactive use of SIGGEN is shown below. Here is a sample session with siggen


% siggen 
Generates a sparse protein signature from an alignment
Domain alignment directories [./]: ../domainalign-keep/daf
Specify mode of signature generation
         1 : Use positions specified in alignment file
         2 : Use a scoring method
         3 : Generate a randomised signature
Select number [1]: 2
Residue contacts scoring method
         1 : Number
         2 : Conservation
         3 : Number and conservation
         4 : None (structural data available)
         5 : None (no structural data available)
Select number [5]: 5
Sequence variability scoring method
         1 : Substitution matrix
         2 : Residue class
         3 : None
Select number [3]: 1
Substitution matrix to be used [EBLOSUM62]: EBLOSUM62
The % sparsity of signature [10]: 15
Window size [0]: 0
Ignore alignment positions with post_similar value of 0 [Y]: Y
Domainatrix signature file output directory [./]: 

Go to the output files for this example




7.0 KNOWN BUGS & WARNINGS

Handling of missing residues in domain alignment files
The alignment in the DAF file (domain alignment file) may be generated by using STAMP via DOMAINALIGN. STAMP will omit from an alignment any residues that either completely lacks electron density and so does not appear in the ATOM records of the PDB file, or which lacks a CA atom. Such residues will of course not be present in the DAF file. This means that acurate gap distances (distance, in residues, between any two residues) for residues from two different alignment positions cannot reliably be found by simply counting residues.

To overcome this problem, data from the domain CCF files (clean coordinate files) are used. These data should be used where available, i.e. the conoption acd option should be set to a value 1, 2, 3 or 4 if possible.

The function embPdbAtomIndexICA is used to create an array which gives the index into the full-length protein sequence for structured residues, i.e. residues for which electron density was determined, EXCLUDING those residues for which CA atoms are missing. The array length is of course equal to the number of structured residues. This array is used for calculating the correct gap distances between residues in the alignment. The domain CCF files MUST be derived from protein CCF files in which residues with a single atom only are ommitted. Such files can be generated by using PDBPARSE with the atommask option set to True. This requirement will not be necessary when a new version of embPdbAtomIndexICA which also excludes residues with a single atom only becomes available.

Manually generated signatures
In the case a signature file is generated by hand, it is essential that the gap data given is listed in order of increasing gap size.

Window size
The user is prompted for a window size that is used for every position in the signature. Likely this is not optimal. A future implementation will provide a range of methods for generating values of window size depending upon the alignment (window size is identified by the WSIZ record in the signature output file).


8.0 NOTES

8.1 GLOSSARY OF FILE TYPES

FILE TYPE FORMAT DESCRIPTION CREATED BY SEE ALSO
Clean coordinate file (for domain) CCF format (EMBL-like). Protein coordinate and derived data for a single domain from SCOP or CATH. The data are 'cleaned-up': self-consistent and error-corrected. DOMAINER Records for residue solvent accessibility and secondary structure are added to the file by using PDBPLUS.
Contact file (intra-chain residue-residue contacts) CON format (EMBL-like.) Intra-chain residue-residue contact data for a protein or a domain from SCOP or CATH. CONTACTS N.A.
Domain alignment file DAF format (CLUSTAL-like). Sequence alignment of domains belonging to the same SCOP or CATH family (or other node in the structural hierarchies). The file is annotated with domain family classification information. DOMAINALIGN (structure-based sequence alignment of domains of known structure). DOMAINALIGN alignments can be extended with sequence relatives (of unknown structure) to the family in question by using SEQALIGN.
Signature file SIG format Contains a sparse sequence signature suitable for use with the SIGSCAN program. Contains a sparse sequence signature. SIGGENLIG, LIBGEN The files are generated by using SIGGEN.
None


9.0 DESCRIPTION

Protein signatures are useful for characterising protein families and have been generated manually in the past (Ison et al, 2000). SIGGEN provides various methods to generate automatically protein signatures.

There are 3 modes for signature generatation:
(1) Use positions specified in alignment file. The alignment file must contain a line beginning with the text 'Positions' for each line of the alignment. A '1' in the 'Positions' line indicates that the signature should include data from the corresponding alignment site. The signature will only include the positions that are marked with a '1'.
(2) Use a scoring method. The alignment is scored (see 'Algorithm') and the signature of a specified sparsity is sampled from high scoring positions.
(3): Generate a randomised signature. A signature of a specified sparsity is sampled at random from the alignment.


10.0 ALGORITHM

Algorithm
signature generation proceeds in three stages as follows: (i) Read data and write residue-residue contact maps. (ii) Apply selected scoring methods to potential signature positions. (iii) Select residues to form the signature and write residue identity and residue gap data into signature output file.

Data Parsing
SIGGEN reads DAF files (domain alignment files) and, optionally, domain CCF files ( clean coordinate files) and CON files (contact files) corresponding to domains in the alignments. If specified, a contact map for each domain in an input alignment is required. A contact map is an N by N matrix (where N is the length of the sequence), a '1' at any element of the matrix indicates contact between the two residues at the corresponding positions, a '0' indicates no contact (see CONTACTS for more information). The data from the DAF files are parsed, including the Post_Similar line (if available, e.g. for DAF files generated by using STAMP via DOMAINALIGN ). The use of the data from the Post_Similar line are fundamental: the user specifies whether only alignment positions with a post_similar value of '1' are considered to be potential signature positions or whether all positions are potential candidates. If the Post_Similar line is not available then all positions are potential candidates. Alignment positions where the Post_Similar value is represented by a '-' are not considered because one or more of the proteins in the alignment were assigned a gap by the STAMP program that was used to generate the alignment.

Residue Scoring Schemes
The algorithm provides four scoring schemes that can be applied to aligned positions (i.e. positions with Post_Similar values that is not '-' or, optionally, '0' either), to enable key residues to be selected for the final signature. The schemes are split into two groups: sequence based and structure based. Each position in the alignment is scored on the basis of a single or combination of 2 scoring schemes, one each from the different groups, thus providing a method of refining/improving the generation of signatures. Every aligned position is allocated a normalised score based on one or more of the following schemes.

Sequence Based Scoring - Residue Identity (ResId)
This scoring function simply takes every residue at a particular aligned position and calculates a score for the substitution of each residue pair using a residue substitution matrix. The average residue substitution score for the position is then normalised and the score assigned to the score array for that alignment position.

eSequence Based Scoring - Residue Variability (ResVar)
This scoring scheme implements the residue variability function of (Mirny & Shakhnovich, 2001).

s(l) = - sum for i=1 to i=6 ( pi(l) x log pi(l) )

Where s(l) is the variability at position l, and pi(l) is the frequency of residues from class i at position l. Six classes of residue are defined which reflect their physical-chemical properties and natural pattern of substitution as follows: (i) Aliphatic (A, V, L, I, M, C); (ii) Aromatic (F, W, Y, H); (iii) Polar (S, T, N, Q); (iv) Basic (K, R); (v) Acidic (D, E); (vi) Special (G, P). The special class represents the special conformational properties of glycine and proline. As a result of this classification mutations within a class are ignored e.g. L to V, whereas mutations that change the residue class are taken into account. Thus each aligned position is given a normalised score that reflects the variability of all the residues in that particular position.

Structure Based Scoring - Number of Residue-Residue Contacts (N-Con)
The contact scoring scheme provides a score based purely on structural information, i.e. the identity and nature of the residues are not considered. The structural information used is the number of residue-residue contacts and the contact maps generated in the first phase of the algorithm are used to derive the number of contacts made by residues at aligned positions. Each residue from an aligned position is noted, and the position that residue occupies in its original protein sequence is determined. The column of the contact map that corresponds to the position of the residue in its original sequence is identified, the occurrence of a '1' anywhere in that column of the matrix is recorded, and the total number of '1's indicates the total number of contacts that residue makes. The number of contacts for each residue at a particular aligned position are determined, the average number of contacts is calculated and the resulting value normalised. This procedure is then repeated for every aligned position.

Structure Based Scoring - Conservation of Residue Contacts (C-Con)
This scoring scheme extends the concept of the number of contacts residues at aligned positions make, by also determining which residues are contacted and their position in the alignment, thus providing a score representing how conserved the contacts made by residues at an aligned position are. The initial stage of the process is identical to that for determining the number of contacts, except every time a contact is found in the contact map, the position of the contacted residue is recorded and its position in the alignment determined. Each residue in an aligned position therefore has associated with it a list of positions in the alignment with which it makes contact. For example if all the residues at position 25 of the alignment make contact with the residues at position 79 of the alignment, a conserved contact is defined and a maximum score is allocated to the residues at position 25. This procedure is repeated for all the contacts made by the residues at position 25 and an average normalised conservation of contact score calculated.

Selection of Signature Positions
The final phase of the algorithm involves selecting the residues that will make up the signature. Following the scoring phase SIGGEN will have created an array of scores for each scoring scheme employed, i.e. a score will have been allocated for every position in the alignment with a Post_Similar value of '1' and optionally '0' also (depending on the Post_Similar option selected, see below). If more than one scoring scheme was used then the scores for each alignment position from the different scoring methods are added together, to give a final array (total score array) of the total scores for each position. It is these final scores that determine which positions will make up the signature.

Signature Sparsity
The signature sparsity is a user-defined parameter that determines how many residues the final signature will contain, for example, if the average sequence length of the proteins in the alignment is 250 residues, then a signature of sparsity 10% (default value) will contain 25 key residues or signature positions, that correspond to the top 25% highest scoring alignment positions.

Key Residue Selection
Assuming that a signature of 10% sparsity is desired and the average sequence length of the proteins is 250 residues, the total score array is re-arranged into ascending order of score. The top (highest scoring) 25 alignment positions (equal to 10% sparsity) are then selected, it is these 25 positions which will make up the final signature. These 25 highest scoring alignment positions are then traced back to the original protein sequences, the residue identities determined and gap data (number of residues between signature positions) calculated. The signature output file is then written, this specifies for each of the 25 signature positions the residues that are observed at that position in the alignment, and the gap (in residues) between that position and the next. In the case of the first signature position the gap data corresponds to the number of residues between the beginning of the sequence and the first position.

Signature Generating Parameters
The SIGGEN algorithm incorporates several options that can be selected when generating a signature. The first is the signature sparsity, which has been introduced above and affects the amount of information encoded in the signature. In addition to the four scoring schemes described above, there are two further option to be considered when generating a signature.

Post_Similar Option
This option determines which alignment positions should be considered as putative signature positions. As mentioned above, the Post_Similar line represents aligned positions by either a '1' a '0' or a '-'. SIGGEN gives the option of considering both positions with values of '1' and '0' or ignoring positions represented by '0', which STAMP considers to be less structurally equivalent, and therefore use just positions with a Post_Similar value of '1'.

Contact Filtering Option
This option also determines which aligned positions should be considered as putative key residues for inclusion in the signature. However, the criterion in this case is whether or not the average number of contacts that the residues at that position make is above a defined threshold (the contact threshold). The default value is 10 contacts, i.e. only aligned positions that make on average 10 or more residue-residue contacts will be considered as potential key residues. As with all the SIGGEN parameters, they can be used in combination. For example, selecting the following parameters: contact threshold = 10; residue identity and conservation of contact scoring schemes; Post_Similar option set to ignore positions with values of '0'; signature sparsity set to 15%, the SIGGEN algorithm would proceed in the following manner: (i) Determine positions with Post_Similar value of '1'; (ii) Determine which of those positions make greater than 10 residue contacts; (iii) Apply the residue id and conservation of contact scoring schemes to the positions resulting from the previous two filtering steps; (iv) Select the top scoring 15% positions to make up the signature. (v) Write signature file.


11.0 RELATED APPLICATIONS

See also

Program name Description
contacts Generate intra-chain CON files from CCF files
domainalign Generate alignments (DAF file) for nodes in a DCF file
domainrep Reorder DCF file to identify representative structures
domainreso Remove low resolution domains from a DCF file
interface Generate inter-chain CON files from CCF files
libgen Generate discriminating elements from alignments
matgen3d Generate a 3D-1D scoring matrix from CCF files
psiphi Calculates phi and psi torsion angles from protein coordinates
rocon Generates a hits file from comparing two DHF files
rocplot Performs ROC analysis on hits files
seqalign Extend alignments (DAF file) with sequences (DHF file)
seqfraggle Removes fragment sequences from DHF files
seqsearch Generate PSI-BLAST hits (DHF file) from a DAF file
seqsort Remove ambiguous classified sequences from DHF files
seqwords Generates DHF files from keyword search of UniProt
siggenlig Generates ligand-binding signatures from a CON file
sigscan Generates hits (DHF file) from a signature search
sigscanlig Searches ligand-signature library and writes hits (LHF file)



12.0 DIAGNOSTIC ERROR MESSAGES

None.


13.0 AUTHORS

Matt Blades

Jon Ison (jison@ebi.ac.uk)
The European Bioinformatics Institute Wellcome Trust Genome Campus Cambridge CB10 1SD UK


14.0 REFERENCES

Please cite the authors and EMBOSS.

Rice P, Longden I and Bleasby A (2000) "EMBOSS - The European Molecular Biology Open Software Suite" Trends in Genetics, 15:276-278.

See also http://emboss.sourceforge.net/

Automatic generation and evaluation of sparse protein signatures for families of protein structural domains. MJ Blades, JC Ison, R Ranasinghe, and JBC Findlay. Protein Science. 2005 (accepted)

A key residues approach to the definition of protein families and analysis of sparse family signatures. JC Ison, AJ Bleasby, MJ Blades, SC Daniel, JH Parish, JBC Findlay. PROTEINS: Structure, Function & Genetics. 2000, 40:330-341

Alignment of a sparse protein signature with protein sequences: application to fold prediction for three small globulins. SC Daniel, JH Parish, JC Ison, MJ Blades & JBC Findlay. FEBS Letters. 1999, 459:349-352.

14.1 Other useful references

LA Mirny EI Shakhnovich. Evolutionary conservation of the folding nucleus. Journal of Molecular Biology (2001) 308:123-129