BACKGROUND: The goal of genomics enabled medicine is to understand how specific molecular pathology of an individual tumor informs therapy tailored to specific vulnerabilities. Exploiting large datasets of human glioblastoma (GBM) and relevant, molecularly-profiled preclinical models enables an iterative learning-loop by which such profiling can produce hypotheses of effective, targeted treatment. Genomic, transcriptomic, and epigenomic profiling of glioblastoma enables depiction of driver-passenger relationships of molecular events among subgroupings of cases of the disease. The present study queries genomic profiles of GBM (TCGA) as well as a panel of patient-derived xenograft (PDX) models. Predictions of targets, and targeted drugs effective against those targets are tested empirically using short-term cultures of GBM PDX models. METHODS: Datamining of TCGA GBM combined with similar genomic profiling of a panel of orthotopic GBM PDX models establishes a data warehouse for discovery of subsets of the disease. These subsets, defined by driver-passenger relationships, partition GBM cases into discreet clusters, which enable follow-on bioinformatic queries for nodes of vulnerability. Short-term cultures of GBM PDX models afford testing cell proliferation, survival, migration, invasion, and differentiation responses to a collection of annotated chemical entities characterized as inhibition a wide spectrum of cancer-related targets (n = 600), establishing a “chemical biology fingerprint” of GBMs of distinct contexts. The chemical biology fingerprints can be mapped against the genomic profiling to find and prioritize candidate therapeutic targets matched to the molecular profile. RESULTS: Human GBM profiling bins tumors into molecular contexts based on driver-passenger relationships; GBM PDX models reside in the same or analogous contexts as primary tumor samples. Within such contexts, biochemical signaling and pathway functions portray associated distinct prioritized candidate targets, which align to varying degrees with Chemical Biology Fingerprints (CBFs). CONCLUSIONS: Multi-omics characterization of human GBM uncovers subclusters of the disease with distinct and unique chemovulnerabilities. Single- and multi-agent studies will afford matching effective interventions tailored to individual tumors. Supported by NIH NCI U01CA168397. SECONDARY CATEGORY: Tumor Biology.