Multiple sequence alignments

Multiple sequence alignments (MSAs) are sets of sequences aligned either to a reference sequence or to each other. They are widely used in bioinformatics:

• Calculate summary statistics such as PSSMs and Potts models
• Infer protein properties, such as structure, via neural networks (MSA Transformer, AlphaFold2/RosettaFold)
• Predict evolutionary relationships (e.g., ancestral sequence reconstruction)

Notes

• The inclusion of MSAs improves zero-shot prediction using PLMs (1) Ref (1)

1.

Su J, Han C, Zhou Y, Shan J, Zhou X, Yuan F. SaProt: Protein Language Modeling with Structure-aware Vocabulary. In: International Conference on Learning Representations. 2024. Available from: https://openreview.net/forum?id=6MRm3G4NiU