Multi-omics reveals shared mitochondrial signatures in cancer, Parkinson’s disease, and heart failure

Introduction: Mitochondrial dysfunction is independently implicated in cancer, neurodegeneration, and heart failure, yet whether these disparate diseases converge on shared mitochondrial targets remains unclear. We performed integrated transcriptomic and proteomic analyses of lung adenocarcinoma (LUAD), Parkinson’s disease (PD), and ischemic cardiomyopathy (ICM) to identify common and disease-specific mitochondrial signatures, using Biomni AI, an AI-based platform, to orchestrate this multi-omics integration. Materials and methods: Differential expression analyses were performed on transcriptomic (GSE140343, GSE114517, GSE116250) and proteomic (IPX0001804000, PXD037684, PXD008934) datasets for LUAD, PD, and ICM, respectively. Gene symbols were HUGO Gene Nomenclature Committee (HGNC)-harmonized, and convergence was assessed via a Venn diagram stratification and Stouffer meta-integration. Pathway-level integration used pre-ranked gene set enrichment analysis (GSEA) against the MitoCarta3.0 collection. Cross-disease analyses combined protein–protein interaction (PPI) network topology, over-representation analyses against a MitoCarta3.0-restricted background, and Drug-Gene Interaction database (DGIdb) drug–gene interaction screening. All integration and enrichment analyses were executed using Biomni AI to demonstrate the feasibility of AI-assisted multi-omics integration. Results: Integrated statistics identified 56 MitoCarta3.0 genes significantly dysregulated across LUAD, PD, and ICM. Sixteen of these were supported by both omics layers in every disease, and an eight-gene core showed transcript–protein directional agreement in every disease. Its four mitoribosomal proteins (MRPL3, MRPL28, MRPS31, MRPS35) were upregulated in LUAD but downregulated in PD and ICM. Oxidative phosphorylation (OXPHOS) components generally trended downward, but only ATP5MF was significantly downregulated in all three diseases. Pre-ranked GSEA identified five pathways (mitochondrial central dogma, mitochondrial ribosome, translation, protein import, sorting and homeostasis, and metabolism) as significantly dysregulated in all three diseases. Network analyses identified mitoribosomal proteins (MRPL13, MRPL3, MRPS31) as the most highly connected proteins. Over-representation analyses of the 56 shared mitochondrial targets against the MitoCarta3.0-restricted background returned no significant terms (0 of 463 tested, false discovery rate (FDR) < 0.05). Drug–gene interaction screening nominated five targets ( COMT , IDH2 , SLC25A5 , RFK , and TXNRD1 ) with cataloged interactions involving Food and Drug Administration (FDA)-approved drugs. Conclusions: Mitochondrial dysfunction in LUAD, PD, and ICM converges on a shared pattern of pathway-level dysregulation spanning the mitochondrial ribosome, translation, and metabolism that is directionally reprogrammed according to proliferative or degenerative tissue states, nominating an eight-gene concordant core and five drug-associated targets as prioritized hypotheses for experimental validation.