“…Eventually 877 high quality samples comprising normal, BPH, tumor, primary, AdPCa, CRPCa, mCRPCa, NEPCa, organoid, PDX (patient-derived xenograft) histology groups were selected and used for HuPSA-based deconvolution analysis. The FASTQ files were collected from multiple cohorts including GSE120795 49 , GSE126078 31 , GSE156289 50 , GSE114740, GSE217260, GSE179321 51 , GSE112786 52 , GSE104131 53 , GSE31528 54 , GSE54460 55 , GSE22260 56 , GSE216490 57 ; dbGaP accessions: phs000909.v1.p1 58 , phs000915.v2.p2 2 . The TCGA_PRAD read counts and TPM data were acquired using “recount3” 59 R package.…”