Developing and Applying RNA Empirical Models With Secondary Structure Insights for Orthoptera Phylogenetics.

The evolutionary patterns exhibited by the ring and stem regions of the RNA secondary structure are distinct. Incorporating RNA secondary structure information into evolutionary models can improve the reliability of phylogenetic trees constructed using RNA sequences. However, empirically derived RNA evolutionary models remain scarce. In this work, we reconstructed conserved secondary structures (MPI > 60%, -score < 0) of 22 mitochondrial tRNAs and two rRNAs in Orthoptera, revealing stronger coevolution and selection signals in paired regions ( value < 0). By encoding sequences into 20-character states and employing maximum likelihood estimation, we developed three substitution models: mtRNA16 (16-state, pairing-specific), mtRNA7 (7-state, uniform mismatches), and mtRNA6 (6-state, mismatch-agnostic). All three models demonstrated robust performance, with the 16-state model providing a better fit to the Orthoptera dataset than the other two models do. Phylogenetic analyses revealed that trees generated using these new models significantly outperformed those generated by a universal model, with the 16-state model producing the most accurate phylogenetic relationships. The results underscore the advantages of incorporating RNA secondary structure information into evolutionary models. In particular, the 16-state model offers enhanced precision in elucidating the phylogenetic relationships in Orthoptera, highlighting its potential for broader applications in RNA-based phylogenetic studies.