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

Dicast achieves 20% higher detection of candidate pathogenic deletions than consensus methods in short-read sequencing data.

• Why it matters: Accurate detection of structural variants is crucial for understanding human diseases, yet current short-read sequencing approaches often miss many variants, limiting clinical diagnostics.
• What they did: The team developed dicast, a machine-learning tool trained on a multi-technology, manually curated ground truth from nine samples, to improve SV call scoring based on alignment and genomic features.
• The result: Dicast outperforms existing methods by identifying more true positives and all pathogenic variants in disease cohorts, enabling more comprehensive and precise genetic diagnoses.