CocoRNA uses cooperative multi-agent RL for RNA design
Original title: CocoRNA: Collective RNA Design with Cooperative Multi-agent Reinforcement Learning
On Rfam, the authors say the agents beat prior methods on success rate and design speed and generalize without extra training.
CocoRNA is a cooperative multi-agent reinforcement learning framework for RNA inverse design. It splits sequence design into smaller sub-problems solved jointly by multiple agents, with a centralized critic using global structural information during training. The authors report that learned policies generalize to unseen structures without further training and outperform prior methods on the Rfam dataset in success rate and design speed. They also report gains on other biological sequence design tasks.
Unconfirmed: Outperformance figures are the authors' experimental claims, not independently verified.
- A centralized critic guides agents with global structural information.
- Policies are said to transfer to unseen structures without retraining.
- Further tests cover other biological sequence design tasks.
From the original article
1 min read at arxiv.orgDesigning RNA sequences that reliably fold into specific secondary structures is essential for understanding their biological functions but remains a challenging computational problem. We propose CocoRNA, a cooperative multi-agent reinforcement learning framework for RNA inverse design. CocoRNA simplifies the design task by decomposing it into smaller sub-problems, each solved collaboratively by multiple agents.
By Tianmeng Hu · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.