Cancer remains a leading cause of death worldwide, with early and accurate diagnosis being paramount for effective treatment. However, traditional diagnostic methods like biopsies are invasive and carry risks, while non-invasive approaches often struggle to identify reliable biomarkers due to tumor heterogeneity and the complexity of biological data. Integrating information across different molecular layers—genomics, transcriptomics, proteomics, and metabolomics—holds promise but is technically challenging, particularly in capturing the dynamic metabolic state of tumors.

Recently, a research team from Yeditepe University, Turkiye, published a groundbreaking article, "Multi-omic data integration and exploiting metabolic models using systems biology approach increase precision in subtyping and early diagnosis of cancer", in Quantitative Biology. To overcome the limitations of isolated data analysis, the team developed an innovative systems biology workflow. They first integrated patient transcriptomic data into human genome-scale metabolic models (GSMMs) to generate personalized metabolic flux distributions (Fluxomic or JX data). These flux profiles, which represent the functional metabolic state of cells, were then combined with genomic, transcriptomic, and proteomic data to train a robust multi-omic machine learning classifier.

This integrated approach was primarily applied to lung cancer, achieving high accuracy in distinguishing cancer from normal tissue, differentiating subtypes (e.g., adenocarcinoma vs. squamous cell carcinoma), and identifying early-stage disease. The classifier pinpointed key biomarker features across all omic layers, including novel metabolic pathways like lipid and amino acid metabolism. Remarkably, the same workflow demonstrated strong performance when applied to pancreatic cancer data, which typically suffers from limited sample availability, proving its robustness and generalizability.


Figure 1 illustrates the core of this integrative systems biology pipeline. It depicts how multi-omic data from sources like TCGA are acquired and processed. Transcriptomic data is fed into a Genome-Scale Metabolic Model to compute patient-specific flux distributions. These flux profiles, alongside genomic and proteomic data, are then used to build a powerful multi-omic classifier. The final output enables precise cancer subtyping, early diagnosis, and the identification of key marker pathways for further investigation.
DOI:10.1002/qub2.70012v