INTREPID: a web server for prediction of functionally important residues by evolutionary analysis

Abstract

We present the INTREPID web server for predicting functionally important residues in proteins. INTREPID has been shown to boost the recall and precision of catalytic residue prediction over other sequence-based methods and can be used to identify other types of functional residues. The web server takes an input protein sequence, gathers homologs, constructs a multiple sequence alignment and phylogenetic tree and finally runs the INTREPID method to assign a score to each position. Residues predicted to be functionally important are displayed on homologous 3D structures (where available), highlighting spatial patterns of conservation at various significance thresholds. The INTREPID web server is available at http://phylogenomics.berkeley.edu/intrepid.

Figure 1.

INTREPID program output. This figure shows the first result returned for input sequence A4UCC6 from the UniProt database. (A) Homologous structure(s) displayed using the Jmol structure viewer. Clicking on the structure pane (Mac: Ctrl-click; PC: Right click) brings up an interactive mode; selecting the console permits users to highlight or modify the display of individual residues or regions in the structure. (B) Users can select different PDB structures for viewing using the pull-down menu; scores displayed are the HMM-based E-values. (C) Users can change the score cutoff used to select residues for highlighting. (D). The pairwise alignment of the selected PDB structure and the query, based on alignment to the HMM constructed for the family. (E) The top 25-ranked residues are displayed. Users can also view all residue scores; score files can be downloaded by following the link at the bottom of the page.

Figure 2.

Result of PhyloBuilder family analysis. (A) The PhyloBuilder ‘book’, showing a variety of data for the family. Top: homologous PFAM domains. Middle: phylogenetic tree, homologous structures, multiple sequence alignment (viewed with Jalview or in hypertext), and a spreadsheet of data. Bottom: Summary information for the family, including GO annotations and evidence codes, taxonomic distribution and predicted subfamilies using the SCI-PHY algorithm. The MSA, phylogenetic tree and HMMs for the family and subfamily can be downloaded from the Downloads link at bottom. (B) The phylogenetic tree for the family, displayed using the PhyloScope viewer. Subtrees of different colors indicate subfamilies predicted by SCI-PHY. Icons are linked to data from external resources: Swiss flags indicate manually curated SwissProt sequences; green flasks indicate experimental support for assigned functions; page icons indicate one or more publications are available. (C) Homologous structures are displayed using Jmol. Positions are colored according to their conservation pattern (e.g. light blue means family wide conservation and dark blue means subfamily specific conservation). (D) Space-fill view of the same structure. Clicking on the structure pane (Mac: Ctrl-click; PC: Right click) brings up an interactive mode, allowing the user to select different display options, including zoom. (E) Plot of conservation patterns at the family (red) and subfamily specific (blue) levels. The alignment of the PDB structure to the family consensus is displayed (derived from aligning the PDB sequence to the HMM for the family).