SNVer

SNVer performs statistical variant calling of common and rare genetic variants from next-generation sequencing (NGS) data, including pooled samples and individual sequences, to detect significant alleles while accounting for sequencing errors.

Key Features:

• Variant scope: Calls both common and rare genetic variants from NGS data.
• Sample types: Supports analysis of pooled samples and individual sequences.
• Statistical framework: Formulates variant detection as a hypothesis-testing problem.
• Model: Implements a binomial-binomial model to test observed allele frequencies against potential sequencing errors.
• P-value output: Produces a single overall P-value for each candidate locus.
• Error-rate control: Allows specification of false-positive error rate thresholds for variant calls to support multiplicity control across loci.
• Decision reporting: Reports statistical significance via P-values rather than solely binary accept/reject decisions.
• Validation: Demonstrated performance on both simulated and real sequencing datasets.
• Scalability: Capable of testing up to 300,000 loci per hour and applicable to whole-exome and whole-genome sequencing analyses on high-performance computing clusters.

Scientific Applications:

• High-throughput variant discovery: Detection of genetic variants in large-scale NGS studies.
• Pooled sequencing analysis: Variant calling in pooled-sample sequencing experiments.
• Rare variant detection: Identification and significance testing of low-frequency alleles.
• Exome and genome studies: Application to whole-exome and whole-genome sequencing datasets.

Methodology:

Variants are called by hypothesis testing using a binomial-binomial model that compares observed allele frequencies to sequencing error expectations and yields a single overall P-value per locus.