Integrative and comparative genomics of tuberculosis disease in humans and livestock

For Chapter 2 of this thesis, which is published in Communications Biology, I analysed peripheral blood (PB) transcriptomics data from 63 control and 60 confirmed M. bovis-infected animals and detected 2,592 differently expressed genes that perturbed multiple immune response pathways. Leveraging imputed genome-wide SNP data, I characterised thousands of cis-eQTL and showed that the PB transcriptome is substantially impacted by intrapopulation genomic variation during M. bovis infection. Integrating the cis-eQTL data with bTB susceptibility GWAS summary statistics, I performed a transcriptome-wide association study (TWAS) and identified 115 functionally relevant genes (including RGS10, GBP4, TREML2, and RELT), which based on their respective expression patterns, stunt the initial proinflammatory immune response during M. bovis infection, suppress the T helper 1 (Th1) T cell response and contribute to macrophage M2 polarisation and Type 2 immunity characteristics that lead to bTB disease susceptibility, bacterial persistence, pathogenesis, and clinical disease. Overall, the outputs from this chapter provide important new omics data for understanding the host response to mycobacterial infections that cause tuberculosis in mammals. For Chapter 3, I adopted a One Health approach and applied the eQTL framework conducted in Chapter 2 to understand the genetic architecture of the human PB transcriptional response during active TB disease and treatment in vivo. Using longitudinal peripheral blood RNA-seq data from n = 48 patients with active TB who underwent anti-TB treatment (ATT), I called sequence variants directly from these transcriptomes and imputed them with a multi-ancestry reference panel. Associating these variants with the expression of nearby genes, I characterised thousands of cis-eQTL and hundreds of response-eQTL (reQTL). I further showed significant changes in cell type proportions during ATT through deconvolution of the bulk RNA-seq data and identified the putative cell type-specific nature of cis-eQTL. Collectively, this work shed light on the immunogenetics of TB disease and treatment, while providing a framework for studies using only RNA-seq data. For Chapter 4, I leveraged the RNA-seq data generated in Chapter 2, in addition to publicly available blood RNA-seq data from cattle naturally or experimentally infected with M. bovis, and cattle infected with other infectious agents (Mycobacterium avium spp. paratuberculosis, bovine herpes virus 1, and bovine respiratory syncytial virus) to investigate the efficacy of peripheral blood mRNA as a host-response biomarker for bTB disease using ML approaches. I identified a 30-gene signature and a 273-gene elastic net classifier that differentiated bTB-positive from bTB-negative cattle, achieving area under the curve (AUC) values of 0.986/0.900 for the former and 0.968/0.938 for the latter in training and testing, respectively. These two classifiers produced high sensitivity and specificity values (≥ 0.853 for both metrics) in the testing set. Additionally, I showed that the two classifiers robustly distinguished bTB+ animals from those infected with other bacterial or viral pathogens achieving AUC values ≥ 0.819. In conclusion, these RNA-based classifiers accurately diagnose bTB and differentiate bTB from other diseases, representing a promising tool for augmenting current diagnostics to advance bTB eradication efforts in endemic regions.