Application of Random Forest and Support Vector Machine on RNA-Seq data to identify immune signatures in H۵N۱ infected chickens

Background: H۵N۱, or highly pathogenic avian influenza, continues to pose a serious risk to public health and poultry. In order to differentiate H۵N۱-infected from healthy chicken macrophages and to guide the development of diagnostics and vaccines, we postulated that machine learning (ML) applied to RNA-Seq data could detect trustworthy immune gene signatures. As far as we are aware, this is the first study to use RNA-Seq data and ML to identify immune signatures of H۵N۱ infection in chickens. Methods: RNA-Seq data from chicken macrophages (ArrayExpress E-MTAB-۲۹۰۸) that were publicly available were preprocessed using low-variance gene filtering and log₂ transformation. Per-gene Z-score normalization was done after features were chosen using an ANOVA F-test (top ۵۰۰ genes; selection was done within each training fold). Fivefold stratified cross-validation (~۸۰/۲۰ train/test per fold) was used to train two supervised machine learning models such as Random Forest and linear-kernel SVM. Performance was evaluated using AUC and ROC curves. Immune-related gene identification was guided by the importance of Random Forest features; a heatmap was used to display the expression patterns of the top genes. Results: Both models achieved robust discrimination between infected and control samples. Random Forest reached mean AUC = ۰.۸۵, while SVM-Linear achieved AUC = ۰.۷۵. Top-ranking interferon-stimulated genes (ISGs) including IFIT۵, MX۱, and OASL were consistently upregulated in infected samples, indicating activation of type I interferon pathways. Concordant findings across models support the stability and biological relevance of the identified signatures despite the modest sample size. Conclusion: The immune signatures of H۵N۱ infection are reliably detected by ML applied to chicken macrophage RNA-Seq. OASL, MX۱, and IFIT۵ are highlighted as potential candidates for early molecular diagnostics and for tracking vaccine-induced antiviral immunity by the Random Forest and SVM-Linear framework, which provides precise classification and interpretable biomarkers. These findings broaden the set of methodological tools available for avian RNA-Seq analysis and could hasten the development of rapid detection systems and next-generation vaccines for poultry health.

کلیدواژه ها:

RNA-Seq ، H۵N۱ ، chicken ، immune response ، machine learning