PASS-based approach to predict HIV-1 reverse transcriptase resistance.

Area of Science:

• Virology and Drug Discovery
• Computational Biology and Bioinformatics
• Molecular Biology and Genetics

Background:

• HIV reverse transcriptase (RT) inhibitors are crucial for managing HIV infection.
• Acquired resistance to HIV RT inhibitors arises from specific mutations in the pol gene.
• Predicting these resistance mutations is vital for effective treatment and drug development.

Purpose of the Study:

• To apply the PASS algorithm for predicting amino acid substitutions linked to HIV-1 RT resistance.
• To leverage publicly available data for computational prediction of drug resistance.
• To assess the utility of a computational approach in guiding anti-HIV drug development and clinical practice.

Main Methods:

• Utilized the PASS algorithm to analyze over 3200 HIV-1 RT variants from the Stanford HIV RT and protease sequence database.
• Focused on amino acid residue and position to establish structure-resistance relationships.
• Employed 20-fold cross-validation for performance evaluation on Phenosense and Antivirogram datasets.

Main Results:

• Achieved an average balanced accuracy of approximately 88% for the Phenosense dataset.
• Obtained an average balanced accuracy of approximately 79% for the Antivirogram dataset.
• Demonstrated the PASS-based algorithm's capability in predicting HIV-1 associated resistance substitutions.

Conclusions:

• The PASS-based computational approach effectively predicts amino acid substitutions conferring HIV-1 resistance.
• This method can assist in selecting appropriate RT inhibitors for HIV-infected patients.
• The predictive tool holds potential for accelerating the development of novel anti-HIV drugs active against resistant strains.