Abstract
Host-microbe interactions are increasingly recognized as an important module to understand disease progression and potential treatment strategies. Increasing evidence points to the microbiome's ability to modulate host gene expression, and thereby influencing host physiology. By integrating dual RNA sequencing with machine learning, we uncover how transcriptionally active microbes (TAMs) may influence host genes involved in immune and metabolic functions in the hospital admitted dengue patients. Towards this, we analyzed 112 whole transcriptomes from the blood samples of patients with differential dengue disease severity. Using a machine learning-based integrated host-microbial transcriptomic analysis framework, combining Lasso regression and sparse canonical correlation analysis (SCCA), we identified both shared and disease-specific associations between the microbes and the host transcriptomic pathways. Notably, opportunistic microbes such as Acetobacter-ghanensis, Achromobacter sp. B7, Bacillus licheniformis, and Clostridium cochlearium, along with the host genes, namely, PPME1, TIMP2, NLRC4, and RhoB, were associated with immune dysregulation in the severe dengue patients. These microbes and genes appear to influence pathophysiology through distinct molecular pathways, highlighting their disease-specific roles in host-microbe interactions.