Differentiating between tumor biopsies and normal cells in the TCGA dataset using targetable cell-surface gene sets

The cancer genome atlas project (TCGA) represents one of the largest sources of mRNA-seq data collected from human tumors. The expression profile of CD, receptor tyrosine kinase, and nuclear hormone receptor transcripts within this data set was investigated using a number of methods including DESeq and t-SNE based clustering. These gene families mediate cell-to-cell communication, adhesion, immune recognition, and hormone response among many roles. By identifying unique expression patterns in TCGA tumor cohorts, mechanisms these cancers use to co-opt host resources, evade the immune system, and proliferate may be identified. Highly expressed members of these cell-surface protein families could also serve as targets for immunotherapies or even small molecule drugs. Here it is shown that TCGA tumor cohorts can be distinguished from TCGA normals using this small, targetable set of 486 genes. Initial insights into the genes within this set that may represent therapeutic targets are discussed.