Processed transcriptomic data and analysis results: cell-type deconvolution of Alzheimer's disease and glioblastoma

Data and code for: Karakuş A. (2026). Apparent shared signature between Alzheimer's disease and glioblastoma reflects converging neuronal loss, not molecular convergence. This deposit contains processed transcriptomic data and analysis results from a cell-type-deconvolution-controlled comparative study of Alzheimer's disease (AD) and glioblastoma multiforme (GBM). The analysis demonstrates that apparent shared transcriptomic signatures between AD and GBM bulk brain tissue (198 shared differentially expressed genes, 96.5% directionally concordant) collapse entirely when cell-type composition is statistically controlled, indicating that observed convergence reflects shared neuronal loss and glial/immune infiltration rather than convergent cell-autonomous molecular biology. Contents:- Normalized expression matrices (GSE48350, GSE36980 RMA-processed; TCGA-GBM + GTEx TOIL TPM)- ComBat-corrected AD meta-cohort matrix with sample metadata- BRETIGEA cell-type composition scores per sample- limma DEG tables (uncorrected and deconvolution-controlled models)- WGCNA co-expression network objects and module assignments- Module hub gene rankings (kME values)- fgsea Hallmark pathway enrichment results- decoupleR/CollecTRI transcription factor activity scores- Grubman et al. 2019 snRNA-seq Seurat object (filtered) Analysis code: https://github.com/akarakus-bartin/Cell-type-deconvolution-analysis-of-AD-vs-GBM-transcriptomics-Karaku-2026 Raw data sources:- GSE48350, GSE36980, GSE138852: NCBI GEO (https://www.ncbi.nlm.nih.gov/geo/)- TCGA-GBM + GTEx: UCSC Xena TOIL pipeline (https://toil.xenahubs.net/)