Computational Drug Repurposing for Gastric Premalignant Lesion Progression: Single-Cell Transcriptomic Identification of SRC/TGFBR2/CAMKK2 Axis Inhibitors in the NAG→CAG→IM→EGC Cascade

Background: Gastric premalignant lesion progression from non-atrophic gastritis through intestinal metaplasia to early gastric cancer lacks approved pharmacological intervention. Methods: We applied a standardized computational drug repurposing pipeline to GSE134520 (Zhang et al. 2019, Nature Communications), comprising 56,440 cells from 13 gastric antral mucosa biopsies (3 NAG control, 10 CAG/IM/EGC disease). After scVI batch correction and Leiden clustering (14 clusters), 13 CASE-enriched clusters (15,834 cells; 13,198 CASE, 2,636 CTL) were subjected to Wilcoxon differential expression (500 DE genes led by REG4, CLDN4, OLFM4, TFF3, FABP1/2). ChEMBL bioactivity screening against 173 druggable targets (20,250 records) and PubChem novelty filtering identified 1,906 novel and 94 known repurposing candidates from the top 2,000 scored compounds. Results: Top novel compound CHEMBL5653589 (score=198.63, N=19 targets, max pChEMBL=8.63) engages TGFBR2, FYN, FASN, HSPA5, and PIK3R1, and has been confirmed as the top-scoring agent across 11 independent disease pipelines in this portfolio. LESTAURTINIB (score=68.52) and MOLIBRESIB (67.80) were the top known candidates. Convergent activity on FYN, TGFBR2, CAMKK2, and MST1R reflects the SRC-TGF-β-Hippo signaling axis governing IM-to-EGC transition. Conclusions: This work identifies 1,906 novel pharmacological candidates for gastric premalignant progression and constitutes the 50th and final provisional patent in the Ritschel Research computational drug repurposing portfolio. Inventors: Glen Charles Ritschel (Ritschel Research, Tega Cay SC) and Claude (Anthropic, San Francisco CA).