Differential Gene Expression Analysis of Latent and Active Tuberculosis used for Machine Learning Guided Therapeutic Peptide Design

Tuberculosis (TB) constitutes a significant and escalating threat to global health. In this study, bioinformatics tools were used to find possible TB hub genes and in silico approaches based on structure and machine learning to target those genes. The study identifies crucial hub genes in three distinct sample types: latent tuberculosis infection (LTBI), active tuberculosis (ATB) and healthy cells, using the GSE62525 dataset from the GEO database. The upregulated genes were used to conduct gene enrichment analysis and construct a protein-protein interaction (PPI) network. Results from the network analysis showed the top ten hub genes. Interleukin-10 (IL10) was identified with promising therapeutic potential in TB. The residue contact was analysed to understand the interaction between the selected crucial node and its receptor. Peptide was built based on the 20:18 residue interface to determine the nature of the interaction between IL10 and its receptor IL-10Rβ. Virtual screening confirmed the stability and interaction of two mutants out of 6,480 mutant peptides that showed significantly increased binding affinities to IL10. Both variant-I (CG_KYC) and variant-II (CV_RYC) peptides exhibited substantial binding to IL10, with variant-II showing the highest affinity, as seen by binding free energies of -68.13 and -95.64 kcal/mol respectively, post-500 ns MD simulation. The study identified active peptides that could lead to future therapies for TB.