On leveraging self-supervised learning for accurate HCV genotyping.

Area of Science:

• Genomics
• Computational Biology
• Virology

Background:

• Hepatitis C virus (HCV) poses a significant global health challenge.
• Current research on HCV primarily uses clinical data, leaving a gap in genomic sequence-based genotyping.
• Accurate HCV genotyping is crucial for effective patient management and treatment strategies.

Purpose of the Study:

• To address the research gap in HCV genotyping using genomic sequences.
• To develop an advanced deep learning approach for accurate HCV genotyping.
• To overcome challenges in computational genomics, such as data scarcity and imbalanced datasets.

Main Methods:

• Utilized Chaos Game Representation for 2D mapping of nucleotide sequences.
• Employed self-supervised learning with a convolutional autoencoder for deep feature extraction.
• Analyzed ten HCV genotypes (1a, 1b, 2a, 2b, 2c, 3a, 3b, 4, 5, and 6).

Main Results:

• Achieved classification accuracy exceeding 99%, outperforming classical and deep learning models.
• Demonstrated effectiveness for both partial and complete HCV genomes.
• Successfully addressed challenges related to imbalanced datasets and data scarcity for certain genotypes.

Conclusions:

• The proposed deep learning model offers a highly accurate and robust method for HCV genotyping.
• This approach provides a valuable benchmark for future HCV genomic studies.
• The model's performance surpasses traditional methods and the NCBI genotyping tool.