Insulin resistance (IR) is the core pathological basis of type 2 diabetes mellitus (T2DM). This study aimed to identify core lipid metabolism-ubiquitination-related feature genes in IR. Integrative transcriptomic and functional analysis screened differential genes via multiple machine learning models. Gene set enrichment analysis (GSEA), regulatory network prediction, and immune infiltration analysis were performed. C2C12 IR models were constructed, and ATG5 was knocked down to detect glucose-lipid metabolism, inflammation, and insulin signaling. Results showed that four feature genes were identified: Good genes Ankyrin Repeat And SOCS Box Containing 4 (ASB4)/ Low Density Lipoprotein Receptor (LDLR) and Bad genes ATG5/ CD36 molecule (CD36 blood group) (CD36). They were enriched in metabolism/inflammation pathways consistent with IR phenotypes. Regulatory network analysis predicted potential upstream microRNAs (miRNAs) and transcription factors (TFs) modulating these feature genes. IR/diabetes groups had lower ImmuneScores vs. insulin-sensitive groups. ATG5 was upregulated in IR models. Its knockdown inhibited autophagy, improved glucose and lipid metabolism, decreased inflammatory cytokines, and exerted no influence on cell viability. In conclusion, ASB4, LDLR, ATG5, and CD36 are potential IR biomarkers. ATG5 regulates IR via autophagy, insulin signaling, and metabolism, serving as a therapeutic target for metabolic diseases.
