RVD

RVD: Hierarchical Bayesian variant detection for low-depth targeted NGS

RVD implements a hierarchical Bayesian model to detect genetic variants and estimate allele frequencies from low-depth targeted next-generation sequencing (NGS) data, enabling accurate variant calling in heterogeneous clinical samples with low sample purity or genetic subpopulations.

Key Features:

• Hierarchical Bayesian Model: Estimates allele frequencies using a hierarchical Bayesian framework to improve robustness in heterogeneous samples.
• Low-Frequency Variant Detection: Achieves high sensitivity and specificity across varying median read depths and minor allele fractions, enabling detection of low-frequency polymorphisms.
• Heterogeneous Sample Analysis: Performs reliable variant calling in samples affected by genomic heterogeneity, including mixed genetic subpopulations.

Scientific Applications:

• Clinical Diagnostics: Identifies mutations in heterogeneous tumor samples, including detection of 15 mutated loci in the PAXP1 gene from a breast ductal carcinoma sample and two likely loss-of-heterozygosity events.
• Cancer Genomics Research: Supports analysis of low-depth sequencing data for studying genetic diseases, tumor heterogeneity, and therapeutic targets.

Methodology:

RVD applies a hierarchical Bayesian statistical model to targeted NGS read data to estimate allele frequencies and classify variants, optimizing performance across varying read depths and minor allele fractions to improve detection accuracy in heterogeneous clinical samples.