Fig. 5: Predicting mutation effects on protein-protein interactions (PPIs).
From: PLM-interact: extending protein language models to predict protein-protein interactions
a Diagrammatic overview of the binary mutation effect classification task. b Inference and fine-tuning of PLM-interact to predict mutation effects that increase or decrease interaction rate/strength. The log formula in this panel is the log-predicted interaction probability ratio between the mutant and canonical pairs. c Precision-Recall curves of two fine-tuned PLM-interact models and four zero-shot models (PLM-interact, TUnA, Topsy-Turvy and D-SCRIPT) on the test dataset. d The ROC curves of the fine-tuned PLM-interact models and four zero-shot models on the test dataset. Source data are provided as a Source Data file.
