lofreq

lofreq calls single nucleotide variants (SNVs) and insertions/deletions (indels) from next-generation sequencing data, modeling base-call quality scores and sequencing-run-specific error rates to detect low-frequency variants while accounting for mapping and alignment uncertainties.

Key Features:

• Error Modeling: Models sequencing-run-specific error rates using base-call quality scores to distinguish true variants from sequencing errors, enabling detection of variants at frequencies below 0.05%.
• High Sensitivity and Specificity: Demonstrates near-perfect specificity and improved sensitivity over existing methods without relying on approximations or heuristics.
• Platform Validation: Validated experimentally on Fluidigm and Sequenom platforms.
• Applicable Organisms: Applicable to viruses, bacteria, and human tumor samples for analysis of genetic variation and heterogeneity.
• Deep Sequencing Efficiency: Efficiently processes deep Illumina sequencing datasets.
• Experimental Validation: Validated on simulated and real datasets from viral, bacterial, and human sources and applied to call rare somatic variants in exome sequencing datasets for gastric cancer.

Scientific Applications:

• Viral and Bacterial Genomics: Identifying rare mutations that contribute to pathogenicity or resistance.
• Cancer Research: Detecting somatic variants in tumor samples to study heterogeneity and evolution.
• Population Genetics: Studying genetic diversity within populations by identifying rare alleles.
• Cell-population Heterogeneity: Leveraging high-coverage sequencing to uncover rare variants within heterogeneous cell populations.
• Clinical Somatic Variant Detection: Calling rare somatic variants in exome sequencing datasets, including studies of gastric cancer.

Methodology:

LoFreq employs a probabilistic framework to model sequencing errors, integrating base-call quality scores and other error sources into its variant calling process.