A Novel Method for Predicting Oncogenic Types of Human Papillomavirus

Area of Science:

• Genomics
• Bioinformatics
• Machine Learning

Background:

• Human Papillomavirus (HPV) causes cervical and anogenital cancers.
• Current HPV classification methods lack scalability and generalizability for emerging types.

Purpose of the Study:

• Develop a machine learning framework for HPV genotype classification by oncogenic risk.
• Introduce TATA-box, CAAT-box, and CpG-island features for HPV risk prediction.

Main Methods:

• Integrated regulatory motifs (TATA-box, CAAT-box, CpG islands) with k-mer composition from HPV genomes.
• Employed multiple machine learning algorithms for performance evaluation.
• Assessed models using accuracy, precision, recall, and F1-score.

Main Results:

• Achieved high classification performance: 97.47% accuracy, 0.95 precision, 0.95 recall, and 0.95 F1-score.
• Demonstrated superior generalizability compared to existing models.
• Validated the effectiveness of the novel feature set.

Conclusions:

• Introduced novel regulatory motif-based features for HPV classification.
• Developed a highly accurate, scalable machine learning model for HPV risk prediction.
• The model shows potential for large-scale screening and vaccine development.