Jace Webster received his B.S. in Bioinformatics and Molecular Biology from Brigham Young University in 2019, where he worked in the lab of Dr. Marc Hansen studying small molecule drug candidates. He then enrolled in the Human and Statistical Genetics doctoral program at Washington University in St. Louis, where he joined the lab of Dr. Christopher Maher for his thesis work. He has been a part of the Maher Lab since 2020 and is primarily interested in the discovery of prognostic biomarkers in cancer and in the development of software tools for cancer genome analysis.
Jace Webster, Ha Dang, Pradeep Chauhan, Alex Shiang, Wenjia Feng, Peter Harris, Russell Pachynski, Aadel Chaudhuri, Christopher Maher
Washington University School of Medicine, St. Louis, MO, USA
Targeted sequencing of cell-free DNA (cfDNA) has emerged as a promising noninvasive method for biomarker detection and disease monitoring. Structural variants (SVs) represent an important class of biomarkers, but detection in cfDNA with limited tumor DNA remains challenging. As a result, detection of both small and large genomic alterations often requires ad hoc combinations of tools not originally designed for cfDNA. To address this, we developed a pipeline integrating standard SNV, indel and copy number workflows with a novel SV detection strategy to allow for sensitive and reproducible analysis. The SV workflow utilizes multiple callers (Lumpy, Delly and Manta) in sensitive mode to nominate candidates, followed by the formation of consensus calls and extensive filtering. This excludes SVs with breakends originating in non-targeted regions or those known to produce high false positive rates. Events supported by matched germline or panel of normal samples are also removed, while those strongly supported in cfDNA (both split-reads and discordant read pairs) are kept. We demonstrated clinical applicability by detecting 100% of published TMPRSS2-ERG fusions and AR duplications, which correlated with overall survival, in a cfDNA prostate cancer cohort. Our pipeline outperforms existing tools (SViCT, Aperture and Factera), showing >50% sensitivity at <=2% tumor DNA abundance in an in silico dilution simulation using data from 4 prostate and 5 colorectal cancer patients, and maintaining the highest specificity using public cfDNA reference data. Our pipeline for reproducible analysis from targeted cfDNA sequencing has potential applications in the noninvasive diagnosis, detection and monitoring of cancer progression.