Abstract
Glioblastoma (GBM) is marked by profound spatial heterogeneity. This complexity fuels tumor progression and therapy resistance. Although recent spatial transcriptomics (ST) studies have defined transcriptional niches, or metaprograms, the systems level regulatory logic linking upstream signaling, transcriptional programs, and therapeutic vulnerabilities remains unresolved. Here, we introduce an integrated, multi layered framework that unifies transcription factor and pathway activity inference (STAN and SPAN) with ligand/receptor and drug target mapping, thereby constructing a systems-level regulatory atlas of GBM across 26 tumors. This atlas uncovers spatial niches organized around transcriptional hubs and signaling programs, including hypoxic mesenchymal regions coordinated by HIF1A and SOX2 with mTORC1 signaling, macrophage rich areas driven by STAT3 mediated immune modulation, and astrocyte-like states shaped by lipid metabolic regulators such as SREBF2. By coupling this regulatory atlas to drug target predictions, we identify niche specific therapeutic vulnerabilities, such as VEGF blockade in HIF1A high regions and MAO-B inhibition in mesenchymal states. Finally, we provide an open, interactive web resource to make these data broadly accessible. This work provides a generalizable blueprint for linking spatial regulation to therapeutic hypotheses across diverse human diseases.