|
The interface surface defined by two or more proteins forming a complex is constructed from polygons in the Voronoi diagram defined by the spheres modeling the proteins’ atoms. The surface satisfies our first intuition, that the interface be about halfway between the proteins. The more difficult question of where to clip the surface is answered by a retraction algorithm following our second intuition, that the boundary be at a relatively tight seal surrounding a wider, more relaxed volume between the proteins. We have created fast software based on a combinatorial algorithm containing ideas on Delaunay triangulations, alpha shapes, and topological persistence. The computed interfaces are visualized geographically, and connected with biological information to allow the biochemist to embed the software within his/her bigger workflow.
Using the basic definition and software, we study the details of interfaces and attempt the relocation of our measurements to observed physical phenomena, such as the importance of residues to the interaction (hot-spots) and the strength or stability of the complexed formation. Although our focus is on proteins interacting with other proteins, the basic ideas are valid for other interactions, like proteins with small ligands, or proteins with RNA, etc. The next steps in the project aim at deepening our understanding of how the observed geometry of the proteins and the interface surface relate to physical properties and how this connection can be wed to the predict function. We plan to approach this question by surveying the known complexes and classifying frequently observed motifs.