PRIDA: Cross-domain protein-RNA interaction prediction based on biological language models and conditional domain-adversarial learning.

PRIDA, a protein–RNA interaction framework using bilinear attention and conditional domain-adversarial learning, improves cross-domain prediction under protein-level distribution shifts and limited labeled data.

• Why it matters: Protein–RNA interaction prediction is constrained by scarce complex structures and experimental labels, which limits supervised models from learning latent patterns needed for RNA-targeted drug discovery.
• What they did: PRIDA leverages pretrained protein and RNA language models, then couples its predictor with conditional domain-adversarial learning; it is further integrated with Monte Carlo Tree Search for virtual screening of RNA aptamer candidates.
• The result: Experiments report state-of-the-art cross-domain performance and show PRIDA can provide effective guidance for RNA aptamer virtual screening for specific target proteins.

AI summary of the abstract · may contain errors · not medical advice · read the paper.