GeoRDe: Tertiary Structure-based RNA Design With Multiple Geometric Constraint
Keywords: RNA design, RNA Inverse Folding, Computational biology
Abstract: Functional RNA sequence design plays an essential role in the regulation of life processes. The RNA inverse folding problem, which involves designing nucleic acid sequences based on their three-dimensional structures, remains highly challenging. This complexity arises not only from the inherent flexibility of RNA structures but also from the base-pairing rules that impose critical spatial constraints on the RNA scaffold. In recent times, the design of RNA has often depended on geometric graph networks to design sequences. Motivated by recent advancements in protein design, we have developed the RNAformer module. This module is capable of learning the geometric constraints of RNA molecules in cooperation with geometric graph networks. Furthermore, to enhance the specificity of sequence generation, we have integrated secondary structure information as labels, ensuring that the designed sequences align more closely with secondary structure constraints. Additionally, we have used RNA language models to understand average evolutionary constraints. By incorporating a range of constraint insights, GeoRDe has demonstrated superior performance under identical training data conditions and has also showcased generalization capabilities on the independent casp15 and RNA-puzzle datasets. Through extensive experimentation, the GeoRDe has proven to be an innovative solution to the challenges of RNA design.
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.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
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: 5731
Loading