The Glioblastoma GBMdrug1000 Dataset Resource Provides Directions for Future Small Molecule Drug Discovery

Abstract Background We previously created a glioblastoma (GBM) DrugBank containing curated information on chemical structure, molecular target activity, and chemical biology for 500 compounds. This study expands the dataset to 1, 103 compounds, including molecular bioactivity, cellular dose-response, CRISPR-Cas9 knockout data, potential toxicity, and links to clinical trials and patents. Methods We gathered information from literature on compounds and models, allowing direct comparisons between compounds, their targets, and biological effects. We also included our own dose-response and drug-induced gene expression data across various glioblastoma cell culture models. Compounds were curated for their effect in preclinical GBM models, and these parameters were projected onto an ECFP₆-based UMAP visualisation. Results The visualisation facilitates comparisons of bioactivities, CRISPR-Cas9 effects in GBM, and potential toxicity in non-transformed models. The analysis highlights the strengths and weaknesses of GBM drug discovery, emphasising the trade-offs between effectiveness, toxicity, and specificity. It also provides insights for optimising targeting based on compound structure and characteristics, targets, and putative toxicity through cheminformatic or experimental approaches. Conclusions The GBMdrug1000 dataset is a public state-of-the-art resource for drug discovery and cheminformatics analysis, complemented by patent information and links to clinical trial data. This curated resource forms a framework for future prioritisation of targets or their combinations.