Multi-omics integration for precision tuberculosis medicine: from molecular atlases to clinical translation

Tuberculosis (TB) remains a major global health threat. Its management is constrained by incomplete prediction of antimicrobial resistance, limited ability to identify patients at risk of anti-tuberculosis drug-induced liver injury (AT-DILI), and heterogeneous treatment responses. This narrative review examines how multi-omics can address these challenges and adapts an established translational framework to TB-specific decisions. Clinically useful integration requires reproducible measurements, independent validation, and evidence that a result improves care. The relevant phenotypes emerge from interactions among Mycobacterium tuberculosis , host genetic and immune variation, lesion microenvironments, drug exposure, and microbial ecology. No single molecular layer adequately resolves this system. We examine pathogen and host genomics, epigenomics, transcriptomics, proteomics, metabolomics, microbiome profiling, and single-cell and spatial technologies. We distinguish replicated resistance and NAT2-associated toxicity findings from emerging bacterial regulatory, immunometabolic, and microbiome associations. Selected diagnostic, prognostic, treatment-monitoring, and toxicity classifiers illustrate both the promise of integration and limitations in sample size, comparator selection, confounding, and validation. Their high training discrimination does not establish external performance or clinical utility. The framework links standardized biobanking and fit-for-purpose integration to analytical and clinical validation, functional testing, and evaluation of utility and equitable implementation. Explicit transition criteria distinguish these stages without treating proposed planning thresholds as universal standards. The clinical value of multi-omics will depend less on the number of measured features than on rigorous study design, affordability, and demonstrable benefit to patients.