Background: Limitations of BCG vaccine in adults and bottlenecks in single-dimensional epitope exploration for tuberculosis R&D.

Existing linear and static analysis guidelines fail to precisely track the complex, multi-stage antigen expression transition process of Mycobacterium tuberculosis during latent and active infection. Classical cell lysis-based analysis or static in vitro peptide screening have critical limitations in reflecting the three-dimensional receptor interactions and inter-species immunological genetic differences within the host's alveolar macrophage microenvironment. This prevents the emulation of in vivo immune feedback fluxes and antigen presentation efficiency within an in silico computational control domain, leading to failures in deriving effective vaccine candidates and ensuring effective immunogenicity and protective efficacy in preclinical stages. Existing pipelines rely on single dominant antigens, making it difficult to acquire cross-protective immunity against the diverse life cycle of Mycobacterium tuberculosis, and face data bottlenecks due to the heterogeneity of host HLA allele genotypes, resulting in inconsistent efficacy across the entire patient population.

Discovery: Multi-Stage Antigen Free-Energy Binding Calculation and Demonstration of Immune Receptor Tensor Synchronization

This study selected nine key immunogens that span the multi-stage expression physiology of Mycobacterium tuberculosis and activated a deep learning-based reverse vaccinology algorithm to integrate and extract CD4+/CD8+ T cell and B cell epitopes. In silico, the interaction with TLR-4 in the alveolar microenvironment and binding free energy were proactively calculated using differential equation-based molecular dynamics simulations, perfectly predicting structural stability and linker/adjuvant binding optimization. This achieves a precision that surpasses existing linear prediction models, revealing the dynamic topological changes of the transcriptome network and maximizing the performance of excluding false-positive epitopes. By synchronizing multi-dimensional, cell-resolution tensor data, batch effects from different datasets were completely eliminated computationally, demonstrating the integrity of the immune response induced by this vaccine platform.

Establishment of a Model for Coordinating Immune Receptor Activation Pathways and Precisely Layering Reversible Immune Homeostasis

Based on the extracted epitope matrix, precise stratification of patients based on family-specific HLA variations was successfully formalized. In particular, a mechanism for up- and down-regulating the rate constants of the TLR-4-mediated downstream signaling pathway was designed to coordinate the rate of antigen presentation while suppressing excessive inflammatory responses in the host's immune system. This establishes a backbone that allows the host's immune homeostasis to be reversibly and autonomously maintained even under metabolic stress and in environments with multidrug-resistant Mycobacterium tuberculosis. This layered model controls the variability of target binding and immune activation profiles across a global population with diverse immunological baselines, providing a foundation for personalized immune simulation companion diagnostics (CDx).

Prospects: Establishing a Standard for Programmable Preventive Immunology and Launching a Next-Generation IND Digital Governance System

This AI-designed, multi-stage mRNA tuberculosis vaccine platform transforms accumulated omics tensors into a programmable vaccine development governance system, resetting the next-generation IND approval process. In Phase III clinical trials or next-generation mRNA clinical scale-up processes led by multinational pharmaceutical companies, a computational barrier has been established to eliminate batch-to-batch variations in immunogenicity by feeding back multi-stage antigen gradient correction coefficients to the model in real-time. This not only meets companion diagnostic (CDx) specifications but also completely solves the limitations of mRNA higher-order structure stability variations that may occur during cGMP large-scale commercial production, becoming a digital master asset that will drastically shorten the timeline for global regulatory approval and clinical entry. This study is expected to establish an innovative backbone platform that will change the paradigm of tuberculosis prevention within the next 10 years.