Effective protein homology modeling requires accurate refinement of loop regions, a puzzle that continues to prove challenging despite recent improvements in sampling algorithms and energy functions.  Extending previous efforts, we approach this puzzle with a test set of antibody partial homology models where five of the six complementarity determining region (CDR) loops are modeled from templates, and the non-CDR-loop scaffold retains the native backbone. We predict the sixth CDR loop, H3, applying the latest sampling algorithms included in the Protein Local Optimization Program, an improved rotamer library, sampling residues surrounding the loop in the inexact homology model environment, and novel methods imposing crystal contacts and docking small molecule antigens into the starting homology model. Preliminary results show an average backbone RMSD of 1.2 � on an initial test set of fourteen 5-8 residue loops.

Predicting H3 loops in antibody homology models

243rd National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.