A Three-View System for CRISPR-Cas12a crRNA Activity Ranking: When to Ensemble, When Context Matters More

TriCasRank improves crRNA activity ranking for CRISPR-Cas12a, raising Spearman correlation from 0.741 (sequence-only) to 0.821 in five-fold target-grouped cross-validation.

• Why it matters: Screening all crRNA candidates experimentally is costly, and existing ranking approaches may miss context captured by reference mapping and historical activity patterns.
• What they did: The authors built TriCasRank by combining sequence features with EasyDesign reference mapping plus activity summaries from related training records, using historical summaries computed within each training fold to prevent information leakage.
• The result: In out-of-fold predictions, TriCasRank’s 0.821 correlation is comparable to a sequence-plus-mapping XGBoost model (SCC 0.822), with larger errors at activity extremes and reduced stability when template identities differ between training and evaluation.

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