Abstract 4280: Improving the sensitivity of wide targeted cancer gene panels via novel genome analysis tools

Abstract

Abstract In the recent years, wide targeted gene panels have gained popularity, due to their ability to interrogate many genes at the same time. In cancer, particularly, there is a set of ~600 genes known as Cancer Gene Census for which mutations have been causally implicated. Many of the wide cancer panels, therefore, target a subset of these genes. As targeted panels widen, the average coverage of them reduces, if the total budget of the DNA reads remains constant. Therefore, wide panels usually have a more modest coverage, as compared to narrow panels. The reduction in coverage, along with the biases that exist in targeted enrichment approaches make a challenging environment for genome assembly pipelines. We argue that, in such systems, false negatives are prone to increase to non-trivial values, primarily due to: 1) inefficiencies in the enrichment methods, especially if they are hybridization-based; 2) sporadic insufficiency of coverage; 3) biases in the analysis pipelines; 4) filtering operations that take place, in attempt to reduce false positives. As much of the losses are due to the dependency of the assays and the pipelines to the reference genome, we have approached this issue by designing a tandem set of tools: 1) A hybrid-denovo genome assembly pipeline, which has significantly less bias to the reference genome, as compared to the conventional align-to-reference methods. This pipeline can be used in lieu of or in combination with the conventional pipelines; 2) An in-silico verification (ISV) tool that provides the experts with a multifaceted visual and textual information, to gauge the evidence (or lack thereof) for the variants of interest. While not meant to necessarily be a replacement to wet validation, ISV provides the ability to identify the errors that may otherwise seep through the system. In order to show feasibility of this approach, a small set of FFPE TruSight Tumor 170 samples including cases of lung and breast cancer was used. The comparisons were done for all RefSeq exonic bases on a hereditary cancer gene subset. Our preliminary analysis using this data shows that by using our proprietary genome analysis pipeline and ISV tool, the sensitivity of the reported variants could be improved from ~80% to ~98% or higher, without impairing the specificity. If we consider intronic bases which have reasonable coverage, including those involved in splicing, the sensitivity improvements could be even larger. Citation Format: Bahram G. Kermani, Evans L. Roberts, Theresa A. Boyle, Anthony M. Magliocco. Improving the sensitivity of wide targeted cancer gene panels via novel genome analysis tools [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4280.