Extracting this archive will create two directories, qptabsearch (about 430 MB when extracted), the top of a hierarchy containing most of the contents, and ptgraph (about 500 KB when extracted) which contains some Python scripts and modules required by others in qptabsearch/scripts/.
Note that some sets of results have been excluded from this archive as they are very large, and this web server has limited disk space and bandwidth. So running "make" in the toplevel directory or the results directory and so on will fail, as there are dependencies on data sets that have not been included.
If you have a 64-bit Linux system, you can simply run the binaries as included, however. If you have a different system or otherwise want to build the software yourself, running make in the qptabsearch/src/ directory will build the software, and running make in the qptabsearch/tests/ directory will build and run the regression tests.
This software can be used freely for any purpose, modified, redistributed, etc. with no restrictions. However we would appreciate it if you acknowledge your use of it, and in particular if you would cite our paper in any publication that makes use of it.
QP Tableau Search is also available for online use via the Pro-origami web server.
Combined tableau and distance matrix database for ASTRAL SCOP 1.73 95% sequence identity subset
Tableau database for ASTRAL SCOP 1.73 95% sequence identity subset
Numeric database for ASTRAL SCOP 1.73 95% sequence identify subset
In the "Non-linear matchings" section of our paper, we describe the use of random permutations of tableaux as an artificial test to verify the capability of our method to find non-linear matchings, i.e. sets of correspondences between SSEs in which the sequential order of corresponding SSEs is not preserved. Here we provide some more details on this method for those who might like to use to evaluate other non-linear structural alignment techniques.
As described in the paper (and citations therein) a tableau is a square symmetric matrix in which each row represents the orientation of one SSE relative to every other SSE. Normally, the rows (and columns) are in the same order as the SSEs in the protein sequence, from N to C terminus. In order to generate randomly permuted tableaux, we instead generate the tableaux by randomly permuting the sequence of SSEs, so that rather than the rows being ordered according to the sequence, they are ordered according to the random permutation that was generated.
The -u option on the pytableaucreate.py script included with the source code is used to perform this task. It also outputs the permutation used, which is necessary so that any alignment based on the permuted tableau can be mapped back to the actual SSEs in the structure which the row and column in the tableau represents. The ssepermutationremap.py script accomplishes this task, and is used by the qptabmatchstructs.sh script when the -u option is used to randomly permute the query structure. This script illustrates the pipeline used to perform an alignment (possibly with a permuted tableau) and visualize it using PyMOL. The genpermutedqueries.sh script was used to generate the queries described in the paper.
All of the scripts mentioned here are provided with the source code download, and all contain internal documentation as to their purpose and usage.
Stivala, A., Wirth, A. and Stuckey, P., Tableau-based protein substructure search using quadratic programming BMC Bioinformatics 2009, 10:153
This page, its contents and style, are the responsibility of the author and do not necessarily represent the views, policies or opinions of The University of Melbourne.