A robust benchmark for detecting low-frequency variants in the HG002 Genome In A Bottle NIST reference material.

Area of Science:

• Genomics
• Human Genetics
• Molecular Biology

Background:

• Somatic mosaicism, where genetic variants arise after conception, is a significant cause of various diseases.
• Detecting low-frequency somatic and mosaic variants is challenging due to their presence in a small fraction of cells.
• Accurate benchmarking is crucial for validating variant callers and understanding mosaic variant allele fractions.

Purpose of the Study:

• To create a benchmark dataset for evaluating the detection of subclonal mosaic variants in normal cell populations.
• To establish a reliable reference material for assessing the accuracy of somatic variant detection technologies.
• To support the optimization and validation of genomic analysis pipelines for mosaic variants.

Main Methods:

• Development of a benchmark using Genome in a Bottle (GIAB) Consortium HG002 reference material DNA from a lymphoblastoid cell line.
• Utilized high-coverage (300x) whole-genome sequencing and a somatic variant caller to identify candidate mosaic variants (>5% allele fraction).
• Validated candidate variants using multiple sequencing technologies (BGI, Element, PacBio HiFi) and defined regions with minimal mosaicism (>2%) for false positive assessment.

Main Results:

• Established a draft mosaic variant benchmark for HG002, including 13 single nucleotide variants (SNVs) in medically relevant genes out of 85 high-confidence candidates.
• Delineated a 2.45 Gbp region within HG002 autosomal benchmarks with no detectable mosaic variants (>2%), facilitating false positive rate evaluation.
• Demonstrated the benchmark's utility in external validation for identifying false negatives and false positives across diverse technologies and algorithms.

Conclusions:

• The developed HG002 mosaic variant benchmark provides a critical resource for the genomic community.
• This benchmark enables robust validation and optimization of methods for detecting low-frequency somatic mosaic variants.
• The findings support the use of this reference material for simulation, spike-in, and mixture studies to improve diagnostic accuracy.