Whole genome and targeted sequencing data have just started to revolutionize clinical care for cancer patients. Based on genetic variants in each tumor, clinicians can tailor treatment options for patients. While genomic sequencing data provide the landscapes of genetic variants, RNA-seq data can provide complementary information and allow us to interpret the functional effects of discovered variants. However, N-of-1 analysis with RNA-seq data poses several challenges, including integration with DNA data and normalization of gene expression. To tackle these challenges, we developed RNA analytics for integrative analysis for tumor samples in a clinical setting.

For variant analysis, we use RNA to validate somatic variants (both SNVs and indels) and to elucidate potential functional effects of the variants. Because RNA-seq has different coverage patterns from WGS, we developed tools to calculate and compare variant allele frequency in both DNA and RNA data, which allows us to validate variants and evaluate allelic expression. Additionally, leveraging on RNA-seq, our tools look for evidence of aberrant splicing events, such as exon skipping and intron retention that are caused by genetic variants. These analyses not only provide an integrative view of the data, but also prioritize variants and potentially reveal variant effects.

Because many cancers are driven by fusion genes and some fusion genes are recurrent in GBM, we use our data to detect gene fusion supported by RNA-Seq, and if possible by WGS. We use FusionCatcher and STAR-Fusion, coupled with OncoFuse to discover and annotate fusions in RNA-Seq. We subsequently look for supporting evidence by any of the structural variant callers we use.

Gene expression quantification is most valuable as a comparative measurement. However, this is challenging for N-of-1 analysis, since such assessment requires proper normalization and comparison to a proper cohort. Batch-effects due to library preparation and other covariates exacerbate the issue of normalization. We tackle this issue with various normalization and correction strategies. Utilizing GC-bias correction, batch-effect adjustment, and public datasets, our analytic provides gene expression measurement that allows researcher to interpret whether the gene is up- or down-regulated compared to a cohort (such as TCGA).

Our RNA analytics comprehensively utilize transcriptomic data and provides solutions to integrative analysis. We have applied our tools to sequence data from five glioblastoma patients and showed how RNA-seq can compliment WGS. Our integrative analysis helps identify an exon-skipping event in MET in one patient, which was validated with Sanger sequencing. The expression analysis also helps identify potential aberrant activity of the SMO pathway, possibly driven by a SMO missense mutation and loss of SUFU.

Citation Format: Bo-Juen R. Chen, Nicolas Robine, Kazimierz Wrzeszczynski, Vladimir Vacic, Anne Katrin Emde, Kanika Arora, Minita Shah, Ewa A. Grabowska, Vanessa Felice, Esra Dikoglu, Catherine Reeves, Mayu Frank, Vaidehi Jobanputra, Michael C. Zody, Toby Bloom, Robert Darnell. RNA-Seq in the NYGC GBM Clinical Outcomes Pilot Study. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4498.