End-to-end learning of multiple sequence alignments with differentiable Smith-Waterman
Preprint
Abstract
Multiple Sequence Alignments (MSAs) of homologous sequences contain information on structural and functional constraints and their evolutionary histories. Despite their importance for many downstream tasks, such as structure prediction, MSA generation is often treated as a separate pre-processing step, without any guidance from the application it will be used for. Here, we implement a smooth and differentiable version of the Smith-Waterman pairwise alignment algorithm that enables jointly learning an MSA and a downstream machine learning system in an end-to-end fashion. To demonstrate its utility, we introduce SMURF (Smooth Markov Unaligned Random Field), a new method that jointly learns an alignment and the parameters of a Markov Random Field for unsupervised contact prediction. We find that SMURF learns MSAs that mildly improve contact prediction on a diverse set of protein and RNA families. As a proof of concept, we demonstrate that by connecting our differentiable alignment module to AlphaFold and maximizing predicted confidence, we can learn MSAs that improve structure predictions over the initial MSAs. Interestingly, the alignments that improve AlphaFold predictions are self-inconsistent and can be viewed as adversarial. This work highlights the potential of differentiable dynamic programming to improve neural network pipelines that rely on an alignment and the potential dangers of relying on black-box methods for optimizing predictions of protein sequences.Keywords
All Related Versions
- Published version: Bioinformatics, 39 (1).
This publication has 40 references indexed in Scilit:
- Assessing the utility of coevolution-based residue–residue contact predictions in a sequence- and structure-rich eraProceedings of the National Academy of Sciences, 2013
- Infernal 1.1: 100-fold faster RNA homology searchesBioinformatics, 2013
- Direct-coupling analysis of residue coevolution captures native contacts across many protein familiesProceedings of the National Academy of Sciences of the United States of America, 2011
- Learning generative models for protein fold familiesProteins: Structure, Function, and Bioinformatics, 2010
- Fast and accurate automatic structure prediction with HHpredProteins: Structure, Function, and Bioinformatics, 2009
- Mutual information without the influence of phylogeny or entropy dramatically improves residue contact predictionBioinformatics, 2007
- Gapped BLAST and PSI-BLAST: a new generation of protein database search programsNucleic Acids Research, 1997
- An improved algorithm for matching biological sequencesJournal of Molecular Biology, 1982
- Identification of common molecular subsequencesJournal of Molecular Biology, 1981
- A general method applicable to the search for similarities in the amino acid sequence of two proteinsJournal of Molecular Biology, 1970