Multiple alignment (PAIRWISE and TREEWISE)

Mammalian and Bacterial Serine Proteinases

(This example is discussed in Russell & Barton, (1992).)

Despite a pronounced functional similarity (a highly conserved catalytic triad), this family of proteins shows little overall sequence similarity. Indeed, sequence alignment methods generally fail to provide an accurate alignment of these protein sequences. In situations like these, STAMP can be used to provide an accuarate alignment of protein sequences based on a comparison of 3D structure. This can often reveal regions of weak sequence similarity that are not detectable during a comparison of sequence. The files for this example are in the directory examples/s_prot in the directory where you have installed STAMP.

A list of the domains is given in the file s_prot.domains. The output. Note that you can create such a file by using the PDBC program. Running PDBSEQ:

pdbseq -f s_prot.domains > s_prot.seqs

produced the file s_prot.seqs, from which an AMPS multiple sequence alignment was produced, and stored in the file s_prot_amps.align. Running ALIGNFIT:

alignfit -f s_prot_amps.align -d s_prot.domains -out s_prot_alignfit.trans

should give the output:

ALIGNFIT R.B. Russell 1995
 Reading in block file...
 Blocfile read: Length: 261
 Reading in coordinate descriptions...
 Reading coordinates...
 Checking for inconsistencies...
 Doing pairwise comparisons...
 Doing treewise comparisons...
 ALIGNFIT done.
 Look in the file s_prot_alignfit.trans for output and details

The final transformation (called alignfit.trans if default ALIGNFIT settings are used) is in the file s_prot_alignfit.trans.

This provides an initial set of transformations for use by STAMP. To run STAMP type:

stamp -l s_prot_alignfit.trans -prefix s_prot

Should produce the following output:

STAMP Structural Alignment of Multiple Proteins
 by Robert B. Russell & Geoffrey J. Barton 
 Please cite PROTEINS, v14, 309-323, 1992


    Sc = STAMP score, RMS = RMS deviation, Align = alignment length
    Len1, Len2 = length of domain, Nfit = residues fitted
    Secs = no. equivalent sec. strucs. Eq = no. equivalent residues
    %I = seq. identity, %S = sec. str. identity
    P(m)  = P value (p=1/10) calculated after Murzin (1993), JMB, 230, 689-694

     No.  Domain1  Domain2  Sc     RMS    Len1 Len2  Align NFit Eq. Secs.   %I   %S    P(m)
Pair   1  4chaa    3est     7.73  10.58    239  240   242  209 207   20  37.08  74.17 0.00e+00 
Pair   2  4chaa    2ptn     7.80   9.43    239  223   234  201 198   20  40.17  75.31 0.00e+00 
Pair   3  4chaa    1ton     6.81   9.38    239  227   245  183 178   19  29.71  66.11 2.19e-42 
Pair   4  4chaa    3rp2a    7.45   9.95    239  224   235  195 188   18  29.71  67.78 2.36e-63 
Pair   5  4chaa    2pkaab   7.27   9.65    239  232   241  198 195   19  29.71  73.64 0.00e+00 
Pair   6  4chaa    1sgt     7.09   9.76    239  223   239  191 183   20  27.62  69.04 3.68e-49 
Pair   7  4chaa    2sga     3.66   9.75    239  181   238  105  99   14  11.72  33.05 2.03e-07 
Pair   8  4chaa    3sgbe    3.56   9.24    239  185   239  103  98   16  10.04  33.47 1.89e-05 
Pair   9  4chaa    2alp     3.49   9.13    239  198   246  102  97   14   9.21  31.38 1.28e-04 

                                      <etc.>
Reading in matrix file s_prot.mat...
Doing cluster analysis...
Cluster: 1 (  2ptn &  2pkaab ) Sc 8.50 RMS 10.20 Len 232 nfit 213 
 See file s_prot.1 for the alignment and transformations
Cluster: 2 (  2sga &  3sgbe ) Sc 8.34 RMS 10.37 Len 191 nfit 164 
 See file s_prot.2 for the alignment and transformations
                             <etc.>    
Cluster: 7 (  2alp &   2sga  3sgbe ) Sc 8.40 RMS 10.30 Len 202 nfit 161 
 See file s_prot.7 for the alignment and transformations
Cluster: 8 ( 1sgt &  4chaa   3est  3rp2a   1ton   2ptn  2pkaab ) Sc 7.59 RMS  9.50 Len 268 nfit 175 
 See file s_prot.8 for the alignment and transformations
Cluster: 9 (  1sgt  4chaa   3est  3rp2a   1ton   2ptn  2pkaab &   2alp   2sga  3sgbe ) Sc 4.77 RMS  9.74 Len 290 nfit 111 
 See file s_prot.9 for the alignment and transformations

The various fields describe details of the pairwise and treewise comparisons: S_c, RMS deviation, the alignment length (Align), the length of each structure in residues (Len1, Len2), the number of atoms used in the RMS fit (Nfit), the number of equivalent secondary structure elements (Secs), and the number of equivalent residues (see above, Eq.).

STAMP will also produce several files:

s_prot.mat - a file containing the information used to derive the structural similarity tree (i.e. the output from the PAIRWISE) mode. This is an upper diagonal matrix containing the pairwise S_c values.

s_prot.N - a series of files containing transformations and alignments created by running the TREEWISE mode in STAMP. Each file corresponds to a node in the similarity tree (i.e. a cluster), where two groups of one or more structures have been combined to form an alignment and transformations. The higher the value of N the more structurally dissimilar the proteins contained in the file are. Highly similar structures are clustered (aligned/superimposed) at an early stage in the program, with more distantly related structures being clustered towards the end.

The top of each file contains the information needed to generate (using TRANSFORM, see below) superimposed coordinates (in STAMP domain format, see below). After these details, various details of the similarity (RMS deviation, S_c value, etc) are given. The bottom portion of the file contains the structural alignment in STAMP format. Positions not aligned with gaps contain information as to the degree of local structural similarity, such as the distance between (averaged) ${\rm C}_{\alpha}$ atoms, and the $P_{ij}^{\prime}$ value.

Methods for displaying sequence alignments and structures are described below.