AGAPE is a prediction-based fold recognition program, that finds remote structural homologues in the PDB database.
Remote homologues (0-25% sequence identity) are detected by a novel prediction-based fold recognition method (Przybylski & Rost 2004). The principle idea is to expand 1-D information of protein sequences by incorporating predicted secondary structure and solvent accessibility states of each amino acid. The resulting 'generalized sequences' are aligned with similarly expanded ('generalized') position specific scoring matrices. The correlation of predicted secondary structure and solvent accessibility states is in most cases higher than between predicted and observed states (Przybylski & Rost 2004). Consequently, AGAPE uses predicted states for PDB proteins in the template library. Alignments produced by AGAPE are on average more similar to structual alignments than the alignments from PSI-BLAST. Regarding pure fold recogniton performance, AGAPE on average improves over PSI-BLAST by about as much as PSI-BLAST improves over BLAST.
Availability and download
This method is currently NOT available for download. Please write to assistant at rostlab dot org.
Commercial licenses can be obtained through Biosof LLC
Dariusz Przybylski & Burkhard Rost