TWAS/PWAS gene expression prediction models

We used the method described by Pain et al. to calculate the gene expression prediction models for QTL summary data. Briefly, this method calculates ten TWAS prediction models using eight leading summary statistic methods (top1, PRS-CS, SBayesR, SBayesR-robust, DBSLMM, LDpred2, lassosum, and SuSiE). Chromosome positions were based on GRCh37 and SNPIDs were based on dbSNP 155. Cis gene regions (upstream and downstream 1 MB for the gene in eQTL and 500 KB for pQTL) were included.brain_PWAS_Wingo et al.brain_PWAS_Yang et al.brain_cell_TWAS_Fujita et al. brain_cell_TWAS_Bryois et al. brain_TWAS_Zeng et al.brain_tissue_TWAS_Klein et al.If employing TWAS or PWAS models in your analysis, please cite: Huang Y.-F., Huang K.-L. (2025) “Single-cell multi‑omic integration analysis prioritizes druggable genes and reveals cell‑type‑specific causal effects in glioblastomagenesis.” medRxiv. DOI: 10.1101/2025.06.28.25330486. ISSN: 3067‑2007.