A Fast Topology-Based Structural Comparison Service

Users are invited to use an updated release of a fast topology-based structural comparison service at http://tops.ebi.ac.uk/tops/compare.html. The service uses TOPS topology diagrams and patterns, and is part of the suite of programs available on the TOPS web-site at http://tops.ebi.ac.uk. TOPS topology diagrams are also linked from the PDB via PDBSum in the "Other Sources" section of the "Structure Explorer" pages.

This service permits comparison of user-supplied target protein domains with the TOPS Atlas (a representative set of 3000.domains), or the entire PDB (over 18,000 domains, as of January 1999). Alternatively, users can supply the descriptions of two protein domains and compare them to each other.

The system also performs an analysis based on possible common motifs contained in the target file, plus some information about strand pairs and secondary structure types. The output per match comprises a distance, and can optionally be annotated with the matching residue numbers, and the common topological pattern.

This structure comparison method is not based on aligning atomic coordinates of residues, but on making a structural alignment of secondary structure elements (helices and strands) using additional information about relationships between strands in sheets, etc., plus chiralities. The method works best for structures containing beta sheets, but an attempt will be made to align all-alpha structures if there is sufficient chirality information.

The advantage of the method is that individual comparisons are fast -- a comparison of one structure against the Atlas takes on average 1-3 minutes of CPU time, and from under 10 minutes to one hour against the entire PDB (on a DEC Alpha). However, other structural comparison techniques based on making structural alignments at the atomic coordinate level are likely to be more accurate.

Uploaded files are deleted after the comparison has been made.

PDB file format: Files will be pre-processed by the DSSP program under Unix. If your input file(s) cannot be processed by DSSP, then the comparison will fail. You may remove any information (e.g., the amino-acid sequence) which will not abort DSSP, and thus preserve some anonymity of your data. You are strongly advised to edit your PDB file so that it only contains the description of the domain in which you are interested. Multi-domain files will take longer to process than single-domain files; also, the comparison databases contain single domain descriptions. Jobs submitted may be subject to queueing in the case of heavy processor usage.

Please mail comments to David Gilbert at drg@ebi.ac.uk, who is responsible for maintaining the service as part of his ongoing research project at EBI as a part-time visiting researcher. The original research effort was supported by an EPSRC grant for DRG as a Visiting Research Fellow at EBI while on sabbatical from City University during 1998.

David Gilbert, drg@ebi.ac.uk
Department of Computing, City University, London, UK and EBI (visiting)

David Westhead, westhead@bmb.leeds.ac.uk
School of Biochemistry and Molecular Biology, Leeds University, UK

Janet Thornton, thornton@biochem.ucl.ac.uk
Department of Biochemistry, University College London, and Crystallography Department, Birkbeck College, and European Bioinformatics Institute