Accurate estimation of mutational effects on protein-protein binding energies is an open problem with applications in structural biology and therapeutic design. Several deep learning predictors for this task have been proposed but, presumably due to the scarcity of binding data, these methods under-perform computationally expensive estimates based on empirical force-fields. In response, we propose a transfer-learning approach that leverages advances in protein sequence modeling and folding stability prediction for this task. The key idea is to parameterize the binding energy as the difference between the folding energy of the protein complex and the sum of the folding energies of its binding partners. We show that using a pre-trained inverse-folding model as a proxy for folding energy provides strong zero-shot performance, and can be fine-tuned with (1) copious folding energy measurements and (2) more limited binding energy measurements.The resulting predictor, StaB-ddG, is the first deep learning predictor to match the accuracy of the state-of-the-art empirical force-field method Flex ddG, while offering an over 10,000x speed-up.

Understanding how changes (mutations) in proteins affect the way they stick to each other is important for developing new medicines and studying biology. Today’s best computer methods for predicting these effects are slow and need a lot of computing power. We developed a new approach using machine learning that learns from many existing protein examples. Our key insight is a way to combine different sources of existing data. Our method is much faster than previous tools and just as accurate. This means scientists can now quickly predict how mutations might affect protein interactions, which could help speed up drug discovery and other research.