West, C. | bioRxiv Science

bioRxiv2018

Integrated Genomic Analysis of Hypoxia Genes across Cancer Types Identifies Significant Associations with Cancer Hallmarks

Hypoxia is a generic micro-environmental factor in most solid tumours. While most published literature focused on in vitro or single tumour type investigations, we carried out the first multi-omics pan cancer analysis of hypoxia with the aim of gaining a comprehensive understanding of its implication in tumour biology. A core set of 52 mRNAs were curated based on experimentally validated hypoxia gene sets from multiple cancer types. The 52 mRNAs collectively stratified high- and low-hypoxia tumours from The Cancer Genome Atlas (TCGA) database (9698 primary tumours) in each of the 32 cancer types available. High- hypoxia tumours had high expression of not only mRNA but also protein and microRNA markers of hypoxia. In a pan cancer transcriptomic analysis, [≥]70% of the known cancer hallmark pathways were enriched in high-hypoxia tumours, most notably epithelial mesenchymal transition potential, proliferation (G2M checkpoint, E2F targets, MYC targets) and immunology response. In a multi-omics analysis, gene expression-determined high- hypoxia tumours had a higher non-silent mutation rate, DNA damage repair deficiency and leukocyte infiltration. The associations largely remained significant after correcting for confounding factors, showing a profound impact of hypoxia in tumour evolution across cancer types. High-hypoxia tumours determined using the core gene set had a poor prognosis in 16/32 cancer types, with statistical significances remaining in five after adjusting for tumour stage and omics biomarkers. In summary, this first comprehensive in vivo map of hypoxia in cancers highlights the importance of this micro-environmental factor in driving tumour progression.

cancer biology↗

bioRxiv2017

sigQC: A procedural approach for standardising the evaluation of gene signatures

With the increase in next generation sequencing generating large amounts of genomic data, gene expression signatures are becoming critically important tools, poised to make a large impact on the diagnosis, management and prognosis for a number of diseases. Increasingly, it is becoming necessary to determine whether a gene expression signature may apply to a dataset, but no standard quality control methodology exists. In this work, we introduce the first protocol, implemented in an R package sigQC, enabling a streamlined methodological and standardised approach for the quality control validation of gene signatures on independent data sets. The emphasis in this work is in showing the critical quality control steps involved in the generation of a clinically and biologically useful, transportable gene signature, including ensuring sufficient expression, variability, and autocorrelation of a signature. We demonstrate the application of the protocol in this work, showing how the outputs created from sigQC may be used for the evaluation of gene signatures on large-scale gene expression data in cancer.

bioinformatics↗