RNA Seq Expression Data Analysis From the Publicly Available Human Gene Samples

Publicly available RNA-Seq from the Cancer Genome Atlas (TCGA) and associated clinical processed data were downloaded from the recount2 project for 319 cases (https://jhubiostatistics.shinyapps.io/recount/) of Caucasian men. Only cases matching primary tumors and within the age threshold (42–58 and 66–73) were retained for analysis. The DESeq2 R/Bioconductor package was used to read and perform analysis of the RNA-seq count data. Differential expression tests were used to compare the differences between young and old patients adjusting for Gleason score (≤3 + 4 and ≥4 + 3). Unsupervised hierarchical clustering was used after a variance stabilizing transformation of the raw count data.

Furthermore, 46 pairs of paired-end RNA-Seq data from TCGA(https://gdc.cancer.gov) corresponding to prostate tumor (n=23) and matching normal tissue (n=23) from 23 patients were selected based on race (Caucasian) and age (young: 42–58, old: 66–73). Briefly, reads were filtered by quality and complexity prior to alignment to human reference genome (hg19) using Star and aggregated by featureCounts. Genes with less than five counts in at least 50% of the samples were filtered out. Differential expression analysis between tumor and normal samples was performed using DESeq2 after stratifying the dataset by age (young and old) and Gleason score (well differentiated: Gleason score ≤3 + 4, poorly differentiated: Gleason score ≥4 + 3). Differentially expressed genes were filtered by false-discovery rate of 0.05 and a log2 fold change less than -1 or greater than 1.