Published: March 01, 2019
Justin M. Zook, Marc L. Salit, Steve Lincoln, Matthew Lebo
The confirmation of genetic variants identified by next-generation sequencing (NGS) using orthogonal assays (e.g., Sanger sequencing) is standard practice in many laboratories. Published studies have examined this issue, concluding that confirmation of the highest-quality NGS calls may not always be necessary. However, these studies are generally small, omit statistical justification, and explore limited aspects of the underlying data. We conducted a rigorous examination of NGS data from two clinical laboratories. Five Genome in a Bottle reference samples and over 80,000 clinical patient specimens were analyzed, with the combined data providing insights that neither data type alone could provide. In total, almost 200,000 variant calls with orthogonal data were examined including 1684 falsepositives detected by confirmation. We used these data to identify criteria that flag 100% of false positives as requiring confirmation while minimizing the number of flagged true positives. Rather than relying on one or two quality metrics, as current publications do, a battery of criteria proved superior, consistent with the metrics recommended by recent practice guidelines. Indeed, our expanded criteria identify some false positives as requiring confirmation that the currently published criteria miss. We also find that historical performance (observing a variant as a true positive some number of times) without other strict quality criteria can also lead to false positives escaping confirmation. Although we found limitations with the currently published criteria, our large, multi-laboratory study reaffirms prior findings that high accuracy variant calls can be separated from those of lower confidence. Our methodology for determining test and laboratory-specific criteria can be generalized into a practical approach that can help reduce the cost and time burden of confirmation without impacting clinical accuracy.
Citation: Journal of Molecular Diagnostics
Pub Type: Journals
genomics, clinical DNA sequencing, validation
Created March 01, 2019, Updated March 15, 2019