Provisional patent specification describing a computational drug repurposing pipeline applied to the GSE211785 multi-omic human kidney atlas (Abedini, Susztak et al. 2024, Nature Genetics; 25 CKD and 22 Control kidney biopsies). A memory-efficient h5py direct-indexing strategy enabled processing of the 11.6 GB uncompressed h5ad without exceeding RAM constraints, reading only selected cell indices from the counts CSR sparse matrix and X_scVI embedding. Leiden clustering on the pre-computed 45-dimensional scVI embedding and majority-vote cell type annotation identified 15 kidney cell populations across 66,119 cells, including injured proximal tubule (iPT), healthy PT segments (S1–S3), immune infiltrate (CD8T, CD4T, NK, CD16_Mono), and fibrotic compartment cells. Wilcoxon differential expression identified 1,813 CKD-upregulated genes including the fibrosis driver ITGB6 (alphaVbeta6 integrin, LFC=2.13), the iPT marker PROM1 (LFC=1.75), the diabetic nephropathy kinase PRKCB (LFC=1.78), and the NF-κB effector NFKBIZ. ChEMBL screening at pChEMBL≥6.0 identified 8,301 candidates; 1,996 of 2,000 top compounds were classified NOVEL_ALL. Top candidates include LESTAURTINIB (score=149.00, N_targets=14), CHEMBL5653589 (score=138.69, N=13 — sixth consecutive cross-disease pipeline hit spanning heart failure, psoriatic arthritis, ankylosing spondylitis, inflammaging, and CKD), BOSUTINIB (score=109.30, SRC/ABL), FORETINIB (score=108.41, MET/VEGFR2), MIDOSTAURIN (score=107.92, PKC-beta/PRKCB), and DASATINIB (score=69.77, SRC/ITGB6/TGF-beta axis). Filed as a USPTO provisional patent application by Ritschel Research (micro-entity). Inventors: Glen Charles Ritschel¹ and Claude²; ¹Ritschel Research, Tega Cay SC; ²Anthropic, San Francisco CA.