UNMASC: tumor-only variant calling with unmatched normal controls

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Somatic mutation detection in cancer is crucial but often requires manual review, especially without matched normal DNA.
• Existing methods for tumor-only variant calling face challenges in distinguishing somatic mutations from germline variants and sequencing errors.

Purpose of the Study:

• To develop and evaluate an accurate and cost-effective computational pipeline for somatic point mutation detection in cancer samples without matched normal DNA.
• To determine the optimal number of normal controls required for reliable tumor-only variant calling.

Main Methods:

• Utilized public databases for known somatic and germline variants.
• Developed the UNMASC (Unmatched Normal-Assisted Somatic Calling) pipeline, leveraging information from nearby positions in normal controls.
• Benchmarked the pipeline using targeted capture panel sequencing data from tumor and normal samples.

Main Results:

• Approximately ten normal controls were sufficient to achieve 94% sensitivity, 99% specificity, and 76% positive predictive value.
• The UNMASC approach significantly outperformed comparable tumor-only variant calling methods.
• The pipeline effectively classifies sequencing errors and private germline variants.

Conclusions:

• The UNMASC pipeline provides a robust solution for somatic variant detection in tumor-only sequencing data.
• This method offers a cost-effective alternative and supplement to traditional matched-normal workflows.
• The approach has potential applications for other next-generation sequencing data analysis challenges.