Here we employed 'pharmacoscopy,' a technology that integrates automated microscopy, drug screening, and machine learning to identify agents affecting cellular states. We conducted ex vivo drug response measurements on peripheral blood mononuclear cells (PBMCs) from 15 healthy donors, testing 588 perturbations, and on primary brain surgery material from 27 glioblastoma patients, testing 132 perturbations. This extensive drug screening revealed corticosteroids as potent deactivators of effector cells like NK and CD8 T-cells, while certain antivirals and antidepressants displayed specific cytotoxic and anti-inflammatory properties, respectively. In primary brain surgery material from glioblastoma patients, we found that specific antidepressants such as Vortioxetine and Fluoxetine had unexpected anti-glioblastoma activity, preferentially targeting neural lineage cells in contrast to immune lineage cells. We also optimized a miniaturized bulk sequencing technique, DRUG-seq, to explore transcriptomic changes post-drug treatment, testing several psychiatric drugs on neural lineage cell lines. This led to the discovery of an immune-like transcriptional signature linked to anti-glioblastoma efficacy in psychiatric drugs and enabled the mapping of functional gene networks. Additionally, we developed machine learning tools to assess cell morphology, cell-cell interactions, and drug target networks relevant to neurological and immunological contexts. We aimed to publish the results of the original research in peer-reviewed journals and we are preparing a manuscript for submission. The current version of the manuscript is on bioRxiv and we have deposited the generated transcriptomic data and drug-target networks in publicly available repositories (GEO) and the lab website that will be made available upon publication of the associated manuscript.
