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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Predicting which proteins interact together from amino-acid sequences is animportant task. We develop a method to pair interacting protein sequences whichleverages the power of protein language models trained on multiple sequencealignments, such as MSA Transformer and the EvoFormer module of AlphaFold. Weformulate the problem of pairing interacting partners among the paralogs of twoprotein families in a differentiable way. We introduce a method called DiffPALMthat solves it by exploiting the ability of MSA Transformer to fill in maskedamino acids in multiple sequence alignments using the surrounding context. MSATransformer encodes coevolution between functionally or structurally coupledamino acids. We show that it captures inter-chain coevolution, while it wastrained on single-chain data, which means that it can be usedout-of-distribution. Relying on MSA Transformer without fine-tuning, DiffPALMoutperforms existing coevolution-based pairing methods on difficult benchmarksof shallow multiple sequence alignments extracted from ubiquitous prokaryoticprotein datasets. It also outperforms an alternative method based on astate-of-the-art protein language model trained on single sequences. Pairedalignments of interacting protein sequences are a crucial ingredient ofsupervised deep learning methods to predict the three-dimensional structure ofprotein complexes. DiffPALM substantially improves the structure prediction ofsome eukaryotic protein complexes by AlphaFold-Multimer, without significantlydeteriorating any of those we tested. It also achieves competitive performancewith using orthology-based pairing.

Pairing interacting protein sequences using masked language modeling