Benchmarking challenging small variants with linked and long reads.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Genome in a Bottle (GiaB) benchmarks are crucial for validating clinical sequencing and developing variant calling methods.
• Short-read sequencing technologies face challenges in accurately mapping complex genomic regions like segmental duplications.
• Existing benchmarks may not fully cover clinically relevant genes or difficult-to-map areas.

Purpose of the Study:

• To expand existing Genome in a Bottle benchmarks using accurate long and linked reads.
• To incorporate challenging genomic regions, including segmental duplications and difficult-to-map areas.
• To improve the accuracy and comprehensiveness of variant calling benchmarks for clinical sequencing.

Main Methods:

• Utilized accurate long and linked reads to generate expanded benchmarks across seven samples.
• Incorporated difficult-to-map regions and segmental duplications into the benchmark datasets.
• Expanded variant sets to include over 300,000 single nucleotide variants (SNVs) and 50,000 insertions/deletions (indels).
• Increased exonic variant representation by 16%, focusing on clinically relevant genes like PMS2.

Main Results:

• The expanded benchmark covers 92% of the autosomal GRCh38 assembly for HG002, excluding problematic regions.
• Identified eight times more false negatives in a short-read variant call set compared to previous benchmarks.
• Added significant numbers of SNVs and indels, enhancing coverage of challenging exonic regions.
• Demonstrated improved identification of false positives and false negatives across different sequencing technologies.

Conclusions:

• The new long-read-based benchmarks provide a more comprehensive and accurate resource for evaluating sequencing pipelines.
• These benchmarks are essential for identifying limitations in short-read sequencing data and improving variant calling algorithms.
• The enhanced datasets facilitate the development of more reliable sequencing methods for clinical applications, particularly for complex genomic regions and genes.