Biological Database Mining for LLM-Driven Alzheimer's Disease Drug Repurposing
INTRODUCTION: The synergy of structured knowledge and large language models (LLMs) may contribute to identifying drugs for Alzheimer disease (AD) drug repurposing (DR). This paper developed a software pipeline that uses LLMs to translate knowledge stored in natural language (such as in scientific texts) to an applicable DR information structure. METHODS: AD-related entries in Gene Ontology and DrugBank were integrated into a Knowledge Graph database to inform LLM prompts. Based on the biological process impact, the LLM provided a suitability rating for DR, taking into account the inhibitory effect of drugs on AD driving processes.. RESULTS: Drugs with a high potential for DR were identified and manually reviewed, also considering adverse effects. Ripretinib and Pertuzumab (both kinase inhibitors) had the highest DR applicable rating across all iterations. DISCUSSION: We propose retrospective analyses, considering the high-rated drugs and their effect on AD patients as a starting point for further (prospective) research.