dicast: a machine learning method for accurate structural variant detection from short-read sequencing data.

Dicast, a machine learning method, detects structural variants from short-read sequencing data with 20% more candidate pathogenic deletions than consensus approaches.

• Why it matters: Accurate detection of structural variants is crucial for understanding human diseases, but current short-read sequencing methods face significant challenges, limiting diagnostic accuracy.
• What they did: The authors developed dicast, trained on a multi-technology ground truth from nine samples with manual curation, and used alignment and genomic-context features to score SV calls.
• The result: Dicast outperforms existing callers and consensus methods, recovering more true positives at high precision and identifying all pathogenic variants in multiple disease cohorts.