Evaluating Discordant Somatic Calls Across Mutation Discovery Approaches to Minimize False-Negative Drug-Resistant...

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Accurate somatic mutation detection via whole-exome sequencing (WES) is critical for personalized cancer treatment strategies.
• The Sequencing Quality Control Phase 2 (SQC Phase 2) project provided a standardized dataset for evaluating analytical validity in WES workflows.
• Variations in library preparation and bioinformatic analysis can significantly impact mutation detection accuracy.

Purpose of the Study:

• To comprehensively evaluate the robustness and analytical validity of various whole-exome sequencing (WES) analysis pipelines for somatic mutation detection.
• To compare the performance of different read aligners and mutation callers, including commercial and open-source options.
• To assess the impact of different enrichment kits on mutation detection rates and identify potential biases.

Main Methods:

• Utilized tumor whole-exome sequencing (WES) data from the US Food and Drug Administration-led Sequencing Quality Control Phase 2 project.
• Evaluated multiple combinations of read aligners (BWA, Bowtie2, DRAGEN-Aligner, DRAGMAP, HISAT2) and mutation callers (Mutect2, TNscope, DRAGEN-Caller, DeepVariant).
• Assessed performance using metrics such as F1 scores and analyzed discrepancies across different DNA enrichment kits and specific mutation calls.

Main Results:

• The DRAGEN pipeline demonstrated superior performance, achieving high mean F1 scores for both single-nucleotide variants (0.966) and insertions/deletions (0.791).
• Among open-source tools, BWA Mutect2 and HISAT2 Mutect2 combinations yielded the best results for single-nucleotide variants (0.949) and insertions/deletions (0.722), respectively.
• Significant discrepancies were observed across enrichment kits, with Integrated DNA Technologies kits showing higher false-negative rates, and specific kits missing key mutations (e.g., CBL, IDH1, PIK3CB).

Conclusions:

• The choice of bioinformatic pipeline and enrichment kit significantly influences somatic mutation detection accuracy in whole-exome sequencing (WES).
• DRAGEN offers high performance, while specific open-source combinations provide robust alternatives.
• Robust bioinformatic analysis is paramount for reliable clinical decision-making in oncology based on WES data.