Abstract 2373: Deep computational analysis of human and mouse specific next-generation sequencing data generated from PDX specimen

Abstract

Abstract Patient derived xenograft (PDX) samples are powerful tools to analyze tumor biology and drug sensitivity of primary tumors in animal models. However, sequencing studies on those samples face the challenge of mouse stromal contamination, which varies in each of the explanted tumor specimen. Thus, without applying mouse and human specific filters on sequencing data, species specific genomic alterations and differences in expression are impossible to detect. To allow mouse and human specific transcriptome/genome analyses, we adapted our NGS data analysis pipelines to filter for mouse and human specific reads. By applying the analysis algorithms on hybrid-capture and transcriptome sequencing data from 60 lung cancer PDX samples, we were able to generate a comprehensive overview on tumor specific alterations, including point mutations, deletions and insertions, copy number alterations, gene fusions and isoform specific expression levels, as well as species specific genomic alterations and transcriptome wide expression levels. From 5781 nonsynonymous substitutions which were detected by hybrid-capture sequencing of 333 cancer relevant genes (CAGE-Rx Scanner), 863 human specific nonsynonymous substitutions were detected after filtering for mouse specific variations. The mean sequencing coverage of the CAGE analysis across all targeted bases in all 60 samples was 662x. An integrated analysis of this data set and corresponding transcriptome sequencing data revealed TP53 and KRAS as the most frequently mutated genes which were expressed. As expected, due to the relative high tumor content in all samples (ranging from 30 to 90% with a median of 90% as determined by pathological review), 80% of all mutations called from transcriptome sequencing in an unfiltered fashion were human specific and called after species specific filtering. In addition, the detection of mouse specific transcriptome sequencing reads identified several growth factors expressed by tumor surrounding tissue, which would have been discarded as unmapped reads in the normal analysis setting. Thus, our analysis alorithms not only enable to remove mouse specific false positive single nucleotide variants (SNVs), but also a detailed analysis of mouse specific expression patterns which might strongly influence tumor growth via the tumor microenvironment. Citation Format: Roopika Menon, Petra Schneider, Martin Peifer, Frauke Leenders, Johannes M. Heuckmann. Deep computational analysis of human and mouse specific next-generation sequencing data generated from PDX specimen. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2373. doi:10.1158/1538-7445.AM2014-2373