Single-cell transcriptomics and machine learning identify macrophage mitochondrial biomarkers for tuberculosis diagnosis
Background: The critical blind spots of existing single/static immune diagnostics and the multi-dimensional host-metabolic omics data bottleneck in tuberculosis R&D.
Existing static analysis standard guidelines, such as sputum smear microscopy or interferon-gamma release assays (IGRAs), have caused serious false-positive results and noise due to cellular dissociation structural collapse in heterogeneous clinical settings, such as immunocompromised patients or children. In particular, bulk transcriptomic analysis techniques fail to capture the intercellular network heterogeneity and dynamic immune-metabolic state changes at the single-cell level within tissue microenvironments, leading to barriers in controlling genetic baseline variations when analyzing biological interactions between macrophages and Mycobacterium tuberculosis. The inability to in silico computationally control the molecular dynamic feedback flux of host immune-metabolic reprogramming due to intracellular colonization of Mycobacterium tuberculosis and chronic resistance feedback flux has resulted in significant omics data bottlenecks in R&D, failing to maintain effective colonization and prophylactic concentrations in the R&D stage, leading to the demise of clinical pipelines.
Discovery: Operation of a multi-algorithm ensemble model and empirical demonstration of single-cell resolution mitochondrial-macrophage gene tensor synchronization.
To disruptively resolve these data bottlenecks, this study utilized single-cell RNA sequencing (scRNA-seq) data to molecularly map the activation and migration signaling network of myeloid immune cells and operated a large-scale consensus machine learning framework equipped with a total of 113 algorithm combinations. This identified six core gene markers (IL1B, ATG3, CYBB, MX1, RPS27A, RPS3) centered on mitochondria-macrophages. The final ensemble model, which organically combines glmBoost and Random Forest, achieved overwhelming diagnostic accuracy exceeding AUC 0.79 in external multi-center cohorts, disruptively surpassing existing simple models. Furthermore, through molecular docking models, the catalytic pocket binding ability of galangin and kaempferol was calculated, and the activation energy was proactively calculated in silico, and the free energy value was adjusted to finally demonstrate the molecular biological integrity of genome tensor synchronization.
Establishment of a macrophage immune-metabolic regulation and reversible mitochondrial homeostasis precision stratification model.
The identified 6-gene signature perfectly stratifies the metabolic reprogramming and intracellular homeostasis control system of macrophages on a multi-dimensional omics matrix backbone. By mapping the expression intensity of CYBB, which generates reactive oxygen species, ATG3, which regulates autophagy, MX1, which dominates the interferon response, and IL1B, which is downstream of the inflammasome, as a multi-dimensional tensor, the study elucidated the metabolic rate-limiting steps at the cellular level for each patient family. By precisely programming and artificially up- or down-regulating these immune-metabolic state rate-limiting constants, a regulatory backbone was established to restore the reversible homeostasis of host cells even under chronic immune evasion and metabolic stress conditions caused by Mycobacterium tuberculosis infection, which serves as a cutting-edge systems biology foundation for providing precision-stratified therapeutic solutions based on patient molecular phenotypes.
Prospects: Establishment of a programmable host transcriptome standard and launch of a next-generation IND digital governance system.
This host-macrophage mitochondrial transcriptome signature completely resets the static, post-hoc symptomatic system of tuberculosis diagnostics and companion diagnostics (CDx) technology into a programmable biomarker infrastructure based on AI-powered multi-dimensional tensor. By linking a unique correction coefficient that can completely eliminate confounding factors and multi-center batch effects that may occur during clinical trial design, the study has secured a unique computational moat that eliminates global clinical batch-to-batch variations. This enables genetic gradient correction in the high-throughput screening stage of large biotech pipelines and facilitates the smooth passage of the FDA and other global regulatory agencies' IND clinical trial protocol approval timeline and cGMP commercial production licensing evaluation framework, thereby becoming a digital governance asset that disruptively shortens the overall licensing timeline.
Tuberculosis (TB) remains a formidable global health threat, yet rapid and accurate diagnostic biomarkers capturing host immune-metabolic dysregulation remain elusive. Here, we aimed to map the TB immune microenvironment and engineer a reliable, explainable diagnostic signature targeting the macrophage-associated immune-metabolic axis. Utilizing single-cell RNA sequencing (scRNA-seq) as an exploratory discovery tool, we initially dissected the intercellular communication network in TB. We then deployed an exhaustive consensus machine learning framework, comprising 113 algorithm combinations, across multiple transcriptomic cohorts to pinpoint core diagnostic features. The optimal model was interpreted using SHapley Additive exPlanations (SHAP) and prospectively evaluated by enzyme-linked immunosorbent assay (ELISA) in an independent pilot clinical cohort. Furthermore, network pharmacology and molecular docking were leveraged to identify potential small-molecule modulators. scRNA-seq analysis highlighted a myeloid-biased immune reprogramming, wherein activated monocytes act as key inflammatory orchestrators through adhesion and migration signaling. Our large-scale machine learning screening identified an optimal glmBoost + RF ensemble model underpinned by a 6-gene mitochondrial-macrophage signature (IL1B, ATG3, CYBB, MX1, RPS27A, RPS3), achieving consistent diagnostic discrimination (AUC > 0.79) across independent cohorts. Pilot clinical ELISA validation confirmed the systemic elevation of IL1B, ATG3, CYBB, and MX1 proteins in TB patients. Furthermore, computational molecular docking models suggested that the candidate phytochemicals galangin and kaempferol exhibit strong theoretical binding affinities within the catalytic pockets of IL1B and ATG3. We derived and provided preliminary validation for an AI-based, explainable 6-gene signature reflecting monocyte immune-metabolic reprogramming in TB. This signature not only demonstrates translational potential as a triage
The host immune-metabolic gene marker discovery in this study goes beyond theoretical exploration of the tuberculosis immune mechanism and directly applies to the actual global companion diagnostics market and the next-generation precision personalized medicine business line.
First, by instantly scanning the macrophage mitochondrial metabolic rate using a multi-algorithm machine learning ensemble scan in the clinical setting, the study eliminates the temporal gap noise of delayed tuberculosis active diagnosis and false positives at the source and protects patient colonization and host immune cell membrane stability.
At the same time, by linking the single-cell transcriptome omics matrix to open-source NCBI GEO and systems biology databases, a companion diagnostics (CDx) panel interface is realized that can virtually simulate false-positive batch effects and genetic gradients during clinical trial design and calculate the effective docking concentration of candidate drug targets IL1B and ATG3 in real time.
Furthermore, when conducting large-scale licensing clinical trials for next-generation host immune-metabolic targeted therapeutics by multinational companies, the study can be used as a backbone infrastructure that maximizes the probability of obtaining clinical trial protocol and cGMP commercial operation licenses from global regulatory agencies by linking the 6-gene expression signature tensor values as correction coefficients to eliminate batch-to-batch variations in patient immune status and multi-center analysis batch variations.