Generative modeling for RNA splicing code predictions and design

Di Wu, Anupama Jha, San Jewell, Natalie Maus, Jacob R. Gardner, Yoseph Barash

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: RNA Splicing, Computational Biology, Transformers, Deep Learning, Bayesian Optimization, RNA Design

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: we construct a transformer based model, TrASPr, which predicts splicing in a tissue specific manner and combine it with a generative bayesian optimization algorithm to design RNA sequence for a tissue specific outcome.

Abstract: Alternative splicing (AS) of pre-mRNA splicing is a highly regulated process with diverse phenotypic effects ranging from changes in AS across tissues to numerous diseases. The ability to predict or manipulate AS has therefore been a long time goal in the RNA field with applications ranging from identifying novel regulatory mechanisms to designing therapeutic targets. Here we take advantage of generative model architectures to address both the prediction and design of RNA splicing condition-specific outcome. First, we construct a predictive model, TrASPr, which combines multiple transformers along with side information to predict splicing in a tissue specific manner. Then, we exploit TrASPr as on Oracle to produce labeled data for a Bayesian Optimization (BO) algorithm with a costume loss function for RNA splicing outcome design. We demonstrate TrASPr significantly outperforms recently published models and that it can identify relevant regulatory features which are also captured by the BO generative process.

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: 4137