Multi-omic data integration and exploiting metabolic models using systems biology approach increase precision in subtyping and early diagnosis of cancer.

Cancer is a complex and heterogeneous disease characterized by various genetic and epigenetic alterations. Early diagnosis, accurate subtyping, and staging are essential for effective, personalized treatment and improved survival rates. Traditional diagnostic methods, such as biopsies, are invasive and carry operational risks that hinder repeated use, underscoring the need for noninvasive and personalized alternatives. In response, this study integrates transcriptomic data into human genome-scale metabolic models (GSMMs) to derive patient-specific flux distributions, which are then combined with genomic, proteomic, and fluxomic (JX) data to develop a robust multi-omic classifier for lung cancer subtyping and early diagnosis. The JX classifier is further enhanced by analyzing heterogeneous datasets from RNA sequencing and microarray analyses derived from both tissue samples and cell culture experiments, thereby enabling the identification of key marker features and enriched pathways such as lipid metabolism and energy production. This integrated approach not only demonstrates high performance in distinguishing lung cancer subtypes and early-stage disease but also proves robust when applied to limited pancreatic cancer data. By linking genotype to phenotype, GSMM-driven flux analysis overcomes challenges related to metabolome data scarcity and platform variability by proposing marker processes and reactions for further investigation, ultimately facilitating noninvasive diagnostics and the identification of actionable biomarkers for targeted therapeutic intervention. These findings offer significant promise for streamlining clinical workflows and enabling personalized therapeutic strategies, and they highlight the potential of our versatile workflow for unveiling novel biomarker landscapes in less studied diseases.