The structure comparison algorithm of Argos & Rossmann
[3], which is the method used by STAMP, requires that the
protein structures being compared are approximately superimposed
initially. If not then structural similarity may be undetected, and
reliable superimpositions and alignments unattainable. This is a very important
thing to remember about STAMP. If initial superimpositions do not yield high
enough scores (i.e. ) or if the structures are generally different,
STAMP will warn you by printing `LOW SCORE' warnings in its output.
The STAMP package provides three methods of arriving at an initial
superimposition. The first of these is to make use of an alignment
derived on the basis of sequence. The program ALIGNFIT requires
that the sequences extracted from the PDB files (using the program
PDBSEQ) are aligned vertically in AMPS block format (see format and
examples below); one can use AMPS or another method of aligning sequences.
The ACONVERT program is included in the distribution to faciliate converting
alignments to AMPS format from other formats; it also possible to use
Jalview (www.jalview.org) to perform the conversion.
STAMP compares all possible pairs of
structures by performing a least squares fit on all equivalenced
atoms. Once all pairwise comparisons are compared, the
program makes use of a tree to superimpose multiply all
coordinates following the tree. Thus the final superimposition
output is the best possible fit of the structure given the
alignment. For an example where ALIGNFIT is used to provide an
initial superimposition, refer to the alignment of the serine
proteinases in Chapter 2.
In instances where multiple sequence alignment is inaccurate,
ALIGNFIT may still be used, though the initial superimpositions may
not be accurate enough for STAMP to find structural similarity.
In such cases, the best way to arrive at initial superimpositions is to use
the SCAN option within STAMP. This option compares a query domain
against a database of target domains and generates a set of
superimpositions of the target domains onto the query domain.
This set of superimpositons constitutes a multiple aligment
that can be used by STAMP as an initial alignmnent. This works particularly well when structures are very diverse. For an
example, see the alignment of the aspartyl proteinase N- and C-
terminal lobes in Chapter 2.