English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Self-supervised neural language models with attention have recently beenapplied to biological sequence data, advancing structure, function andmutational effect prediction. Some protein language models, including MSATransformer and AlphaFold's EvoFormer, take multiple sequence alignments (MSAs)of evolutionarily related proteins as inputs. Simple combinations of MSATransformer's row attentions have led to state-of-the-art unsupervisedstructural contact prediction. We demonstrate that similarly simple, anduniversal, combinations of MSA Transformer's column attentions stronglycorrelate with Hamming distances between sequences in MSAs. Therefore,MSA-based language models encode detailed phylogenetic relationships. Wefurther show that these models can separate coevolutionary signals encodingfunctional and structural constraints from phylogenetic correlations reflectinghistorical contingency. To assess this, we generate synthetic MSAs, eitherwithout or with phylogeny, from Potts models trained on natural MSAs. We findthat unsupervised contact prediction is substantially more resilient tophylogenetic noise when using MSA Transformer versus inferred Potts models.

Protein language models trained on multiple sequence alignments learn phylogenetic relationships