Zhang et al. carried out proteomic analyses of 95 human colorectal cancer (CRC) samples previously characterized at the genomic level by The Cancer Genome Atlas (TCGA). This proteogenomic analysis found that mRNA levels often did not reliably predict protein levels. Therefore, potential driver genes could be prioritized as those with greater changes in protein levels. CRCs were classified into five proteomic subtypes; these were similar to the three transcriptomic subtypes identified by TCGA but also identified additional characteristics that might better explain CRC biology and patient prognosis.