Computational protocol: IDH mutation status is associated with a distinct hypoxia/angiogenesis transcriptome signature which is non-invasively predictable with rCBV imaging in human glioma

Similar protocols

Protocol publication

[…] RNA sequencing raw data (mapped to genes) and curated IDH-1/2 mutation data were downloaded from TCGA data portal on October 15th, 2014. In total, RNA sequencing and curated mutation data was available for 288 unique samples with low-grade diffuse (WHO °II) or anaplastic (WHO °III) gliomas (detailed sample information is provided in ) which were then analyzed using Bioconductor 3.0, a package implemented in R 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria). First, normalization and differential gene expression analysis of RNA sequencing counts was performed using the edgeR package, which assumes a negative binomial distribution of count data, filtering lowly expressed transcripts.To study relative pathway activity on the level of individual samples, we used the GSVA package. Using a nonparametric approach, GSVA transforms a gene-by-sample gene expression matrix into a gene set–by-sample gene set enrichment matrix, facilitating the identification of differentially activated genesets for each sample. GSVA was performed on 17 gene sets related to hypoxia, vasculo- and angiogenesis (a detailed listing is provided in ) identified through the Molecular Signatures Database (MSigDB - www.broadinstitute.org/gsea/msigdb) and literature research. Next, empirical Bayes moderated t statistics were calculated, and gene sets with FDR-q < 0.05 were considered to have significantly differential activity between samples with mutant and wild-type IDH-1/2 tumors. Of these gene sets, individual GSVA enrichment scores were plotted as a heatmap with the gplots package.Furthermore, the differential upstream biological causes and probable downstream effects between mutant and wild-type IDH-1/2 tumors were analyzed with the proprietary Ingenuity Pathway Analysis (IPA, QIAGEN, Redwood City, CA, USA) using a fold change filter of |1.5| and FDR-q < 0.05. Briefly, the software calculates both an overlap p value (based on Fisher’s exact test) and an activation z score, which is based on the expression state of activating and inhibiting genes, for manually curated upstream regulators and downstream biological functions. Only results with both an FDR-q < 0.05 and a z score > |2| were considered significant. […]

Pipeline specifications

Software tools edgeR, GSVA, GSEA, gplots, IPA
Databases TCGA Data Portal MSigDB
Application RNA-seq analysis
Organisms Homo sapiens
Diseases Brain Neoplasms, Glioma, Neoplasms