Predictive analysis for pathogenicity classification of H5Nx avian influenza strains using machine learning...

Area of Science:

• Virology
• Computational Biology
• Machine Learning Applications in Animal Health

Background:

• Avian influenza (AI) outbreaks cause significant economic and livestock losses globally.
• Determining the pathogenicity of H5Nx strains is crucial for disease control and zoonotic risk assessment.
• Traditional methods for assessing AI pathogenicity often rely on identifying specific markers in the hemagglutinin (HA) gene.

Purpose of the Study:

• To evaluate the predictive performance of various machine learning (ML) techniques for in-silico pathogenicity prediction of H5Nx avian influenza viruses in poultry.
• To utilize complete HA gene sequences for classifying H5Nx viruses as highly pathogenic (HP) or low pathogenic (LP).
• To compare the effectiveness of different ML classifiers in accurately predicting AI pathogenicity.

Main Methods:

• Annotation of 2137 H5Nx HA gene sequences based on the presence of the polybasic HA cleavage site (HACS).
• Comparison of ML classifiers including logistic regression (LR), random forest (RF), K-nearest neighbor (KNN), Naïve Bayes (NB), support vector machine (SVM), and convolutional neural network (CNN).
• 10-fold cross-validation technique applied to raw and aligned nucleotide (DNA) and protein sequences for pathogenicity classification.

Main Results:

• Machine learning techniques achieved high classification accuracies (approximately 99%) for predicting H5Nx avian influenza pathogenicity.
• Classifiers like LR (L1/L2), KNN, SVM (RBF), and CNN demonstrated superior performance on aligned DNA and protein sequences, reaching accuracies of ~99.20%.
• CNN also showed strong performance on unaligned sequences, achieving accuracies of up to 99.20%.

Conclusions:

• Machine learning methods are effective tools for the regular classification of H5Nx virus pathogenicity in poultry.
• These predictive models can assist experts in rapidly determining the pathogenicity of circulating AI viruses, especially when characteristic markers are present.
• The study highlights the potential of ML for enhancing AI surveillance and informing biosecurity measures to mitigate economic and health impacts.