Figure 12. Diagram summarizing our new general strategy for the discovery of companion diagnostics. This simplified strategy can be potentially applied to any cancer type. The first step involves the generation of a novel cellular model, which is interrogated by proteomics analysis. Then, these results are used to establish the prognostic value of these candidate biomarkers, by searching pre-existing human transcriptional profiling data, linked to clinical outcome (in silico validation). The prognostic value of these biomarkers can also be enhanced significantly, by using more than one marker in combination, forming a short signature. This “Proteomics-to-Genomics (PTG)” approach then efficiently yields new targets and biomarkers, linked to parameters associated with clinical outcome (tumor recurrence, distant metastasis, overall survival, or response to therapy).