RNAJog: Fast Multi-objective RNA Optimization with Autoregressive Reinforcement Learning



Introduction

Codon optimization is essential in mRNA vaccine development, while existing tools face limitations in computational efficiency, sequence diversity, and universality. To address these issues, we developed RNAJog (RNA Joint Optimization with Generative model), a framework combining autoregressive generation and reinforcement learning to optimize codon sequences for minimum free energy (MFE) and codon adaptation index (CAI). Evaluations in both dry and wet experiments have confirmed RNAJog's effectiveness and efficiency.

Availability: The RNAJog is available at http://www.csbio.sjtu.edu.cn/bioinf2/RNAJog/ and the source codes of RNAJog are available at https://github.com/kxstd/RNAJog.




Figure 1. The flowchart of the proposed RNAJog  


The output of RNAJog is provided as a CSV file that includes the optimized RNA sequence, MFE (Minimum Free Energy), CAI (Codon Adaptation Index), and other configuration parameters. For more detailed guidance on how to use the tool, please refer to the help documentation

© 2017 Computational Systems Biology/Shen Group.