How AlphaFold and related models predict protein-peptide complex structures

Protein-peptide interactions mediate many biological processes, and access to accurate structural models, through experimental determination or reliable computational prediction, is essential for understanding protein function and designing novel protein-protein interactions. AlphaFold2-Multimer (AF2-Multimer), AlphaFold3 (AF3), and related models such as Boltz-1 and Chai-1 are state-of-the-art protein structure predictors that successfully predict protein-peptide complex structures. Using a dataset of experimentally resolved protein-peptide structures, we analyzed the performance of these four structure prediction models to understand how they work. We found evidence of bias for previously seen structures, suggesting that models may struggle to generalize to novel target proteins or binding sites. We probed how models use the protein and peptide multiple sequence alignments (MSAs), which are often shallow or of poor quality for peptide sequences. We found weak evidence that models use coevolutionary information from paired MSAs and found that both the target and peptide unpaired MSAs contribute to performance. Our work highlights the promise of deep learning for peptide docking and the importance of diverse representation of interface geometries in the training data for optimal prediction performance.