Protinfo FSSA - functional signatures from structural alignments

The meta functional signature (MFS) module has superseded this module and may be more useful for general function prediction; however this module is still extremely useful for the specific task for which it was designed.

Wang K, Samudrala R. FSSA: A novel method for identifying functional signatures from structural alignments. Bioinformatics 21: 2969-2977, 2005.
Wang K, Samudrala R. Automated functional classification of experimental and predicted protein structures. BMC Bioinformatics 7: 278, 2006.

Name:
Email:
Structure:
SCOP fold:

Given a protein structure in PDB format and the SCOP fold family that it belongs to (sample file from the alpha/alpha toroida (a.102) fold family), the FSSA server will return a prediction of the SCOP superfamily that it belongs to along with its functional signature. The submitted structure file should contain only one chain and one structural domain.

The training structure data for the FSSA server is retrieved from the ASTRAL database version 1.67 with less than 95% pairwise sequence identity. Currently the server can classify protein structures for 42 SCOP folds that contain sufficient number of structures and contain at least two superfamilies. About 39% structures in the ASTRAL database belong to these 42 folds. When the sequence identity between the query sequence and all database sequences is less than 30%, the FSSA method has better predictive power than global sequence and structure comparison methods. We however urge biologists using this for function prediction to take all inputs into account along with any available experimental information.

The FSSA software is also available as a standalone application.

Protinfo || Bioverse || Samudrala Computational Biology Research Group || protinfo@compbio.org