Peptone team presents a new method for estimating the change in stability resulting from point mutations at the NeurIPS 2024 Conference

Our AI team, led by Carlo Fisicaro and Michele Invernizzi, is presenting ProteinMPNN-ddG, a new method for estimating the change in stability resulting from point mutations in both monomeric and multimeric proteins.

We have enhanced ProteinMPNN from the Institute for Protein Design, University of Washington, through an additional term derived from the model itself, where only backbone atoms of the single residue being predicted are given, without sequence or structural context, analogously to standard classical free energy calculations. Additionally, we have introduced a new tied decoding scheme, reducing the time complexity from O(N^2) to O(NlogN) and incorporating a full sequence context.

This speedup enables the identification of mutations affecting protein stability at the proteome scale.

Our implementation, combined with a high-performing parser parallelized on NVIDIA GPUs, makes the package incredibly fast.