SMaSH

SMaSH evaluates germline variant calling algorithms by generating synthetic datasets and standardized metrics to benchmark accuracy and computational performance for detection of single-nucleotide polymorphisms (SNPs), insertions and deletions (indels), and structural variants from next-generation sequencing (NGS) data.

Key Features:

• Synthetic datasets: Generates synthetic datasets that mimic real-world genetic variation to provide a controlled environment for benchmarking variant callers.
• Benchmarking on real genomes: Organizes and interprets existing benchmarking data from real genomes to assess algorithm performance across diverse genomic contexts.
• Accuracy and computational metrics: Proposes standardized metrics for evaluating both accuracy and computational efficiency of variant calling methods.
• Variant-type coverage: Evaluates performance across single-nucleotide polymorphisms (SNPs), insertions and deletions (indels), and structural variants.
• Assessment of existing algorithms: Benchmarks a wide range of existing germline variant calling algorithms to reveal strengths and weaknesses.

Scientific Applications:

• Algorithm development and validation: Uses synthetic datasets and standardized metrics to develop, tune, and validate germline variant calling algorithms.
• Comparative benchmarking: Enables objective comparison and ranking of variant calling methods across datasets and variant types.
• Research and clinical evaluation: Supports assessment of variant detection accuracy for genomic research and clinical interpretation using NGS data.

Methodology:

Generates synthetic datasets that simulate genetic variation, benchmarks a wide range of existing variant calling algorithms against those datasets and benchmarking data from real genomes, and applies proposed metrics for accuracy and computational performance.