Fig. 1

The workflow for data pre-processing, model fitting and functional analysis. The flowchart illustrates the working process of the present paper. RNA-Seq and clinical survival data were retrieved from Broad GDAC firehose. mRNA expression data from Illumina Hiseq were used. The RSEM-derived TPM were log2 transformed and standardized. Genes with median absolute deviation (MAD) greater than zero were fused with clinical survival data. In the model fitting section, the derived data were directly applied to the log-rank test or were examined for the proportional hazards assumption before applying the Cox model. Both models were fitted individually for each gene in each cancer type. The result tables indicate the simplified information generating from the models. Multiple testing corrections were performed before subsequently analysed by pathway enrichment and clustering. Abbreviation: RSEM, RNA-Seq by expectation maximization. RMST, restricted mean survival time. TPM, transcript per million