BLAST & FastA
Database Searching Using Short Peptide Sequences is an excellent short article by Bruno A. Gaëta posted on the ABRF web site.
Gaëta explains that, for a peptide sequence of 8 to 10 residues, a standard BLAST search may not report any matching sequences because the matches which have been found are not regarded as being statistically significant. This behaviour can be modified by either increasing the "Expect" parameter to its maximum or by decreasing the "Z" parameter. Another useful tip is watch out for the low-complexity filter, which may need to be turned off if the amino acid composition of the query peptide is unusual.
For short stretches of amino acid sequence, NCBI advise choosing either the identity matrix or a low PAM number (pam40 or pam30).
For a FastA search, the critical parameter is the "Cutoff expectation". This may need to be increased from its default value when the query sequence is short.
The following gateways provide access to both BLAST and FastA unless marked otherwise.
- NCSA Biology Workbench
- NCBI (BLAST only)
- Baylor College of Medicine
- Indiana University (FastA only)
For a comprehensive listing of online search engines, see the page maintained by ABIM.