Integrative Transcriptomic Analysis Reveals Distinct and Shared Host Responses in Dengue and Chikungunya Infections.

Dengue virus (DENV) and chikungunya virus (CHIKV) co-circulate in many regions and present with overlapping clinical features, which complicate accurate diagnosis and disease management. This study develops an integrative transcriptomic framework to identify robust host gene signatures that distinguish between dengue, chikungunya, and healthy states. Publicly available RNA sequencing (RNA-seq) datasets derived from human blood samples were analyzed using a cross-validation design to ensure robustness and prevent information leakage. Differential expression analysis was performed independently within each dataset using the Generalized Linear Models with Quasi-Likelihood F-tests and Magnitude-Altitude Scoring (GLMQL-MAS) framework, followed by Cross-Magnitude-Altitude Scoring (Cross-MAS) integration to identify shared and virus-specific gene signatures. A strict consensus approach across folds was applied to derive reproducible gene sets. These signatures were used for dimensionality reduction and multinomial logistic regression to evaluate classification performance. A small subset of selected genes showed strong discriminative performance within the cross-validation framework, with test balanced accuracy reaching 0.97, which improved upon models using all genes. Biologically, both infections exhibited a shared antiviral response characterized by interferon signaling and innate immune activation. However, distinct virus-specific patterns were identified. Dengue infection was associated with cell-cycle and DNA replication pathways, while chikungunya infection showed stronger enrichment of inflammatory and immune signaling pathways, including NF-kappaB and Toll-like receptor signaling. Overall, this study provides a cross-validation-based framework for integrative transcriptomic analysis and identifies compact, reproducible host-response signatures with strong discriminative signals in the analyzed cohorts. These signatures require validation in larger independent cohorts before any clinical or diagnostic application.