Abstract
The prediction of protein-protein interfaces requires both the
identification of interface residues and the proper spatial orientation
of the component proteins. Many methods have been developed to identify
interface residues, often using relative interface propensity (RIP), the
enrichment of a particular amino acid type at the interface compared to
the rest of the protein surface. We aimed to improve RIP for interface
identification by incorporating the solvent accessibility of each amino
acid. We studied the surface residues of 290 unbound structures
corresponding to components of protein complexes and compared the
relative solvent accessible surface area (rSASA) distributions of
residues that end up in the interface and those that do not. Our results
show that the side-chains of amino acids that become interface residues
are more solvent exposed than non-interface surface residues on the
unbound protein structure. Using this knowledge, we created an
rSASA-dependent probability of becoming an interface residue for each
amino acid type. Our results show that the solvent accessible surface
area of residues should be taken into account when identifying interface
residues and can be applied to other interface prediction techniques
that use RIP to improve their results.