gatk_variant_recalibrator

gatk_variant_recalibrator recalibrates variant calls in Variant Call Format (VCF) files from sequencing technologies by learning a Gaussian mixture model over variant annotations and assigning log-odds (LOD) scores that estimate the likelihood each variant is true.

Key Features:

• Input format: Accepts Variant Call Format (VCF) files containing initial variant calls from sequencing technologies.
• Annotation-based modeling: Learns a Gaussian mixture model over variant annotations to model distributions of true variants and sequencing artifacts.
• Per-variant scoring: Assigns informative log-odds (LOD) scores to each variant that reflect the estimated probability of being a true variant.
• Variant evaluation and refinement: Evaluates and refines variant calls using the learned model to distinguish true variants from artifacts.

Scientific Applications:

• Variant calling accuracy: Reduces false positives and false negatives in genomic variant call sets through recalibration.
• Genomic research: Applicable to large-scale studies including cancer genomics, population genetics, and personalized medicine.

Methodology:

Accepts VCF files; learns a Gaussian mixture model from variant annotations; evaluates each variant using the learned model and assigns LOD scores.