2021 FDA Science Forum
HIVE Dragen Pipeline Enables Somatic Tumor-Only Filtration
- Authors:
- Center:
-
Contributing OfficeCenter for Biologics Evaluation and Research
Abstract
Background
Identifying somatic mutations is an essential step in cancer studies and diagnosing, treating, and monitoring cancer. A matched normal (germline) sample may not be available due to common circumstances, such as financial considerations or retrospective studies. In such cases, challenges exist to identify somatic variants in tumor-only next-generation sequencing (NGS) samples.
Objectives
To develop a filtering pipeline in the High-performance Integrated Virtual Environment (HIVE) to identify somatic SNV and indel variants in tumor-only sequencing samples.
Methods
Publicly available whole genome sequencing data from a triple-negative breast cancer cell line (HCC1395) was processed through the HIVE-adopted Illumina DRAGEN small variant pipeline. The filtration method utilized variant features and population variant databases to filter germline variants in Dragen VCF files and retain somatic variants. Results were benchmarked against a somatic truth set for HCC1395.
Results
Raw DRAGEN output contained >100,000 small variants. Benchmarking of the tumor-only filtration method, using whole genome sequencing and a somatic truth set for HCC1395, identified 843 of 1160 somatic variants (72% sensitivity) while only 117 “false positive” variant calls (0.12 false discovery rate) remained .
Conclusion
Population variant databases have become increasingly comprehensive enabling effective germline filtration strategies for tumor-only sequencing samples.
