Researchers at USTC release open-source RNA foundational model
Scientists at the University of Science and Technology of China have developed structRFM, a model designed to improve the prediction of RNA structures and functional analysis.
Researchers from the MIRACLE laboratory at the University of Science and Technology of China (USTC) have launched structRFM, an open-source foundational model for RNA sequence analysis. The findings, published in the journal Nature, aim to advance artificial intelligence applications in biological research and drug discovery.
The model was developed by integrating structural knowledge into large-scale sequence learning. By utilizing approximately 21 million computed sequence-secondary structure pairs, the team created a system that can predict structural conformations and identify key biological features, such as splice sites, more accurately than existing tools.
To encourage further scientific collaboration, the team has released the complete dataset, model weights, and source code. This open-access approach is intended to provide the scientific community with standardized resources for RNA folding and functional inference tasks.