Input form

The job submission starting page is located at http://www.expasy.org/tools/scanprosite/ on the ExPASy web server (7).

Here the user has to enter sequence and/or signature data, choose the output format, and specify various scan behavior options before launching the analysis job. The form usage is described at http://www.expasy.org/tools/scanprosite/scanprosite-doc.html.

User sequences and/or UniProtKB sequences/databases can be scanned against user patterns and/or PROSITE signatures/database. Multiple sequences and/or signatures can be submitted at once (maximum: 8 for scans against databases, otherwise 16). In this case, any specified signature will be searched against any specified sequence (logical OR).

For scans against protein databases, the search space can be reduced to specific taxa, and/or to entries containing a specific term in their description (UniProtKB DE line).

Protein databases can be randomized (reversed or shuffled) to evaluate user pattern significance (8). Pattern matching mode can be altered (see aforementioned online documentation page under ‘pattern matching mode’).

Users can choose to retrieve full sequences (in fasta format) of matched proteins together with signature match results.

For scans against the whole PROSITE signature database, ‘unspecific’ signatures with a high probability of occurrence (see PROSITE user manual at http://www.expasy.org/prosite/prosuser.html) can be excluded (this option is activated by default). Scans can also be restricted to patterns.

Users can also choose to include low level profile matches (with score smaller than the high confidence - level 0 - cut-off) in the output (this option is not activated by default).

To save bandwidth and computing power, there is a limit on the number of distinct matching sequences and total matches that can be displayed. Users can select the maximum of matching sequences (by default, not more than 5000 matches in 1000 sequences will be displayed). A maximum of 10 000 or 100 000 sequences (50 000 or 500 000 signature matches) can only be requested in the non-interactive ‘batch mode’ where results are sent to the user via email.

The non-interactive ‘batch mode’ is used automatically when an email address is specified.

Instructions for simple programmatic access (text/plain output) are available on demand (email: gro.ysapxe@etisorp).

Sequence analysis

Once the job has been submitted, the data are posted to an application that will make use of pre-computed matches whenever possible. Results are retrieved from an internal relational database that stores matches of all PROSITE signatures (except the ‘unspecific’ signatures with a high probability of occurrence) against the UniProt Knowledgebase, including additional splice variants, and PDB (9). For analyses where no pre-computed results are available (e.g. on user sequences, user patterns or ‘unspecific’ signatures) a real-time sequence analysis is performed using the ps_scan program (10). Users requiring high-throughput sequence analysis or who wish to use custom databases can download ps_scan stand-alone tool at ftp://ftp.expasy.org/databases/prosite/tools/ps_scan/.

Available results are pooled and saved (for 12 h) on the server, for later use by the rich viewer.

The detection of intra-domain features is performed on the fly by the rich viewer over the matching domains (and is therefore only available in the rich view output mode, see below). Profile match regions are evaluated for fulfillment of feature detection conditions specified in the associated ProRules. The features themselves are represented in the ProRule as UniProtKB/Swiss-Prot annotation blocks that are applied when detection conditions are fulfilled. The feature annotation blocks and their detection conditions are retrieved from an internal database storing a relational representation of the ProRules.

Detection conditions can be: specific amino acids inside the domain (regular expressions that can be grouped by logical operators), groups of conditions in which all conditions must be fulfilled (e.g. catalytic triad of trypsin protease).

To date, intra-domain feature detection is performed for 136 of the 595 PROSITE profiles (PROSITE release 19.20, of 07-Feb-2006). Predicted features include (Table 1): binding sites for chemical groups (such as Heme, ATP), glycosylation site, disulfide bridges, DNA binding sites, metal ion binding sites, post-translational modification sites (such as phosphotyrosine, phosphoserine) etc. The system can be easily extended to use other UniProtKB/Swiss-Prot feature types.

The authors would like to thank Eric Jain for careful proof-reading. This work was supported by grant no. 3152A0-103922/1 from the Swiss National Science Foundation and by the Swiss Federal Government through the Federal Office of Education and Science. Funding to pay the Open Access publication charges for this article was provided by Swiss Institute of Bioinformatics (Swiss-Prot group), Switzerland.

Conflict of interest statement. None declared.