SMASH

SMASH identifies sample swaps in high-throughput sequencing datasets by comparing genotypes at approximately 6,000 genome-wide single nucleotide polymorphisms using a Bayesian framework to match samples across sequencing modalities.

Key Features:

• Extensive SNP Integration: SMASH uses approximately 6,000 single nucleotide polymorphisms distributed throughout the human genome to perform genotype-based sample matching.
• Cross-Data Type Verification: It compares genotypes across RNA-Seq, exome sequencing, and MethylCap-Seq data to verify sample identity between sequencing modalities.
• Robust Bayesian Framework: SMASH employs a Bayesian framework that integrates evidence across SNPs to infer sample matches while accommodating variations in data quality and coverage.
• Validation Across Diverse Data Sets: The method has been validated across multiple sequencing datasets to establish identity between different sequencing modalities.
• Performance with Low-Quality Data: SMASH can identify matching samples with low read coverage, with reported performance down to about twenty million reads for relatively low-quality RNA-Seq samples.

Scientific Applications:

• Sample swap detection in large-scale omics studies: Identify and correct inadvertent sample swaps to protect data integrity in medium to large-scale sequencing projects.
• Multi-modal data integration and verification: Verify sample identity across multi-omics datasets to support combined analyses of RNA-Seq, exome, and MethylCap-Seq data.
• Quality assurance for sequencing projects: Provide genotype-based verification to improve robustness of downstream biological interpretation in genomics studies.

Methodology:

SMASH compares genotypes at approximately 6,000 genome-wide SNPs and applies a Bayesian framework to integrate evidence across SNPs and across sequencing modalities (RNA-Seq, exome sequencing, MethylCap-Seq) while accounting for variation in data quality and coverage.