VariantTools

VariantTools performs detection, filtering, calling, comparison, and plotting of genetic variants to identify low-frequency variants in heterogeneous samples such as tumors.

Key Features:

• Variant Detection and Calling: Implements algorithms capable of detecting variant allele frequencies ranging from 0.05 to 1.0, enabling analysis of mixed genetic backgrounds and tumor heterogeneity.
• Reference Dataset Utilization: Leverages a reference dataset from whole-genome sequencing of Coriell cell lines NA19240 and NA12878 (maternal lineages of YRI and CEU HapMap trios) mixed in proportions 10Y/90C, 50Y/50C, and 90Y/10C to simulate genetic heterogeneity.
• High-Throughput Sequencing: Analyzes datasets with ~1.4 billion reads per mixture and average coverage of 64X to provide comprehensive genomic assessment.
• Performance Evaluation and Comparison: Enables benchmarking of variant calling algorithms against established pipelines such as GATK, reporting metrics including a 0.028 false discovery rate (FDR) and 0.019 false negative rate (FNR) for the 50Y/50C mixture at coverage >20X.
• Discovery of Novel Variants: Identified over 475,000 variants not present in published genotypes, with 97% validation accuracy in coding regions.

Scientific Applications:

• Cancer Genomics: Identification of low-frequency somatic variants and assessment of intratumoral heterogeneity.
• Personalized Medicine: Discovery of patient-relevant variants that may inform targeted treatment decisions.
• Tumor Evolution and Clonal Diversity: Characterization of clonal composition and evolutionary dynamics in heterogeneous tumor samples.
• Variant Discovery and Benchmarking: Discovery of novel variants and quantitative benchmarking of variant callers using controlled mixed-sample datasets.

Methodology:

Performs variant detection, filtering, variant calling, comparison (including benchmarking versus GATK), and plotting using whole-genome sequencing data from mixed NA19240 and NA12878 samples.