Cancer precision medicine requires clinically actionable biomarkers for patient stratification and a better prediction of clinical outcome. Although thousands of cancer-enriched mutated genes have been reported by global sequencing projects, to date, only a few oncogenic mutations have been confirmed as effective biomarkers in cancer therapies. The low frequency and varied profile (i.e., allele frequency, mutation position) of mutant genes among cancer types limit the utility of predictive biomarkers. The recent explosion of cancer transcriptome and phenotypic screening data provides another opportunity for finding transcript-level biomarkers and targets, thus overcoming the limitation of cancer mutation analyses. Technological developments enable the rapid and extensive discovery of potential target-biomarker combinations from large-scale transcriptome-level screening combined with physiologically relevant phenotypic assays. Here, we summarized recent progress as well as discussed the outlook of transcriptome-oriented data mining strategies and phenotypic assays for the identification of non-genetic biomarkers and targets in cancer drug discovery.