MSA Generation with Seqs2Seqs Pretraining: Advancing Protein Structure Predictions
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: Protein Language Model, Protein Structure Prediction
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
TL;DR: We introduce MSA-Generator, a self-supervised model that generates virtual MSAs, enhancing protein structure predictions in key benchmarks.
Abstract: Deep learning, epitomized by models like AlphaFold2 \citep{jumper2021highly}, has achieved unparalleled accuracy in protein structure prediction. However, the depth of multiple sequence alignment (MSA) remains a bottleneck, especially for proteins lacking extensive homologous families. Addressing this, we present \METHODNAME{}, a self-supervised generative protein language model, pre-trained on a sequence\textbf{s}-to-sequence\textbf{s} task with an automatically constructed dataset. Equipped with protein-specific attention mechanisms, \METHODNAME{} harnesses large-scale protein databases to generate virtual, informative MSAs, enriching subpar MSAs and amplifying prediction accuracy. Our experiments with CASP14 and CASP15 benchmarks showcase marked LDDT improvements, especially for challenging sequences, enhancing both AlphaFold2 and RoseTTAFold's performance.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3959
Loading