Deciphering Autoantigen Signatures in Chikungunya Virus Infection Using Machine Learning: A Data-Driven Approach to Understand Host Immunity.

Abstract

Chikungunya virus (CHIKV), a mosquito-borne alphavirus, causes acute febrile illness and often progresses to chronic arthritis-like symptoms for which the underlying molecular mechanisms remain elusive. This study identifies key biomarkers of CHIKV-host interactions, shedding light on potential mechanisms underlying virus-induced joint pathology. RNA sequencing data from peripheral blood samples of paediatric patients with natural Chikungunya infection (15-17 days post-symptom onset; GSE99992: Severe cases = 42, Non-Severe cases = 44) was analysed using binary classification models with StratifiedKFold validation, ensuring a robust and reliable approach to feature selection. A panel of 20 gene features selected by recursive feature elimination with cross-validation (RFECV) showed overlap with known autoantigens and were differentially expressed in CHIKV infection. Network analysis revealed interactions among host biomarkers-THG1L, SLC44A5, KCNN3-and viral components such as nsp4 (CHIKV RNA polymerase) and BCL2-like 11 (an apoptosis facilitator), highlighting a multifactorial virus-host interplay. Fibronectin 1 (FN1) emerged as a central hub gene, known for its role in osteoblast mineralization, skeletal development and its association with renal pathologies. These findings provide novel insights into CHIKV-induced immune dysregulation and offer a foundation for future experimental validation and therapeutic exploration.