Improved protein complex prediction with AlphaFold-multimer by denoising the MSA profile

Abstract

Structure prediction of protein complexes has improved significantly with AlphaFold2 and AlphaFold-multimer (AFM), but only 60% of dimers are accurately predicted. A way to improve the predictions is to inject noise to generate more diverse predictions. However, thousands of predictions are needed to obtain a few that are accurate in difficult cases. Here, we learn a bias to the MSA representation that improves the predictions by performing gradient descent through the AFM network. We effectively denoise the MSA profile, similar to how a blurry image would be sharpened. We demonstrate the performance on seven difficult targets from CASP15 and increase the average MMscore to 0.76 compared to 0.63 with AFM. We evaluate the procedure on 334 protein complexes where AFM fails and demonstrate an increased success rate (MMscore>0.75) of 8% on these hard targets. Our protocol, AFProfile, provides a way to direct predictions towards a defined target function guided by the MSA. We expect gradient descent over the MSA to be useful for different tasks, such as generating alternative conformations. AFProfile is freely available at: https://github.com/patrickbryant1/AFProfile