Improving sequence alignments with AlphaFold2 regardless of structural modeling accuracy

This is the authors' abstract. We don't add key points for this paper.

Protein sequence alignments are essential to structural, evolutionary and functional analysis but their accuracy is often limited by sequence similarity unless molecular structures are available. Protein structures predicted at experimental grade accuracy, as achieved by AlphaFold2, could therefore have a major impact on sequence analysis. Here, we find that multiple sequence alignments estimated on AlphaFold2 predictions are almost as accurate as alignments estimated on experimental structures and significantly superior to sequence-based alignments. We also show that AlphaFold2 structural models of relatively low quality can be used to obtain highly accurate alignments. These results suggest that, besides structure modeling, AlphaFold2 encodes higher-order dependencies that can be exploited for sequence analysis.