SPOT-peptide identifies peptide-binding proteins (PBPs) and peptide-binding sites based on structural similarity to known peptide-binding templates.
The query structure is initially superimposed against all proteins in a library of PBPs.
The quality of the superposition is evaluated by SP-score.
Model complex structures between the query protein and template peptide may be derived from top scoring superpositions.
Potential PBPs are subsequently filtered according to 2 measures of local quality:
- A measure of binding site sequence similarity is evaluated by calculating the mean BLOSUM62 substitution score between aligned residues <4.5Å from the superimposed template peptide (EVO).
- The local consistency is also evaluated based on the DFIRE binding interface energy between the query structure and the superimposed template peptide (DFIRE).
Finally, queries with at least one template with SP-score >0.66 after filtering are predicted to be peptide-binding. Remaining queries are classified as non-binding.
Query: The name of the query structure.
Sites: The number of non-redundant predicted binding sites for the query protein (MCC<0.3).
SP-score: The measure of domain-level structural similarity between the query and top-scoring template.
TPL: The PDB chain ID of the top-scoring template protein.
TPL-peptide: The BioLiP ID of the peptide associated with the top-scoring template.
Cluster size: The number of redundant templates which imply the diven predicted binding site.
DFIRE: The interface energy between the query protein and superimposed template peptide normalised by peptide length (lower better).
EVO: Mean sequence similarity between query-template aligned residues in the putative binding region (higher better).
Residues: The predicted binding residues (based on original query PDB numbering).