RoseTTAFold2-PPI: Predicting Protein Interactions with AI
Researchers at Qian Cong’s Lab have released RoseTTAFold2-PPI, a deep-learning model designed to predict how proteins interact with each other. Built on the RoseTTAFold2 architecture, the tool enables large-scale screening of protein-protein interactions (PPIs), a fundamental task for understanding biology and drug discovery.
Proteins rarely function alone; they form complexes that drive nearly every process in the cell. Identifying which proteins bind together has traditionally required time-consuming experiments. RoseTTAFold2-PPI speeds up this process by using paired multiple-sequence alignments and structural information to estimate the likelihood of protein-protein interactions.
The model outputs residue-level contact probabilities, offering detailed insight into where and how two proteins might bind. Because it’s optimized for high-throughput predictions, it can analyze thousands of protein pairs quickly, making it valuable for mapping cellular networks or identifying new therapeutic targets.