Abstract: Objective To explore the biomarkers of cuproptosis in diabetes mellitus (DM) for drug screening. 
Methods The GEO database was used to obtain GSE25724 dataset of DM as a training set and screen its differentially expressed genes, and 19 cuproptosis-related genes were obtained from the literature. The intersection of the two was used to obtain cuproptosis-related genes in DM, which were then validated (GSE23343 and GSE20966 were used as a test set). Gene correlation and enrichment analyses were performed on cuproptosis-related genes in DM. Molecular docking of key genes with predicted compounds was performed using DS BIOVIA Discovery Studio 2016 software. 
Results A total of six diabetes-associated cuproptosis-related genes (DBT, DLD, GLS, PDHB, NFE2L2, and LIPT1) were obtained, and they were involved in biological processes such as metabolic catabolism of organic acids, synthesis of pyruvate-acetyl-coenzyme A and amino acid metabolism, mainly through the regulation of the signaling pathways of the tricarboxylic acid (TCA) cycle of lipidation, pyruvate metabolism, glycolytic glucose de novo, and HIF-1. Among these six genes, DLD, DBT and NFE2L2 were significantly differentially expressed in the test set samples and had good diagnostic value. Molecular docking showed that DLD and DBT could have a high binding capacity with folic acid, NFE2L2 with Butin, and NFE2L2 with Luteolin. 
Conclusion Cuproptosis-related genes DLD, DBT, and NFE2L2 play an important role in the occurrence and development of DM, and this study provides data support for the studies of the pathogenesis and treatment of DM. 

Key words: diabetes mellitus, cuproptosis, computational biology