xTrimoDock: Rigid Protein Docking via Cross-Modal Representation Learning and Spectral Algorithm

Abstract

Protein-protein interactions are the basis for the formation of protein complexes which are essential for almost all cellular processes. Knowledge of the structures of protein complexes is of major importance for understanding the biological function of these protein-protein interactions and designing protein drugs. Here we address the problem of rigid protein docking which assumes no deformation of the involved proteins during interactions. We develop a method called, xTrimoDock, which leverages a cross-modal representation learning to predict the protein distance map, and then uses a spectral initialization and gradient descent to obtain the roto-translation transformation for docking. We show that, on antibody heavy-chain and light-chain docking, and antibody-antigen docking, xTrimoDock consistently outperforms the state-of-the-art such as AlphaFold-Multimer and HDock, and can lead to as much as a 10% improvement in DockQ metric. xTrimoDock has been applied as a useful tool in protein drug design at BioMap.

xTrimoPGLM: Unified 100B-Scale Pre-trained Transformer for Deciphering the Language of Protein

“…However, xTrimoPGLM-AbFold, which does not use any MSA or template information, performs comparable with AlphaFold-Multimer, indicating that xTrimoPGLM-Ab-1B has learned sufficient and rich information on antibodies. Crucially, xTrimoPGLM-AbFold achieves a speedup of 6,300 × over the original AlphaFold-Multimer and 103 × over the faster MSA-searching AlphaFold-Multimer ( 120 ), owing to the original AlphaFold-Multimer consumes a long time to search MSA (0.8 hours per sample). When we increase the number of Evoformer blocks to 16, xTrimoPGLM-AbFold attains the best performance on all metrics while still maintaining a 2,400 × speedup than the original AlphaFold-Multimer and 40 × speedup than the accelerated AlphaFold-Multimer.…”