Comparative study of methods for detecting sequence compartmentalization in human immunodeficiency virus type 1.

Area of Science:

• Virology
• Evolutionary Biology
• Computational Biology

Background:

• Human immunodeficiency virus (HIV) infects various organs, leading to distinct viral subpopulations.
• Restricted viral trafficking can cause these subpopulations to evolve independently, a phenomenon known as viral compartmentalization.
• Compartmentalization is typically identified through comparative sequence analysis, particularly in tissues like the central nervous system and genital tract.

Purpose of the Study:

• To systematically compare the effectiveness of six commonly used methods for detecting HIV sequence compartmentalization.
• To evaluate method performance across diverse datasets, including clinical samples and simulated sequences.
• To provide recommendations for reliable inference of compartmentalized population structures.

Main Methods:

• Comparative sequence analysis of HIV-1.
• Systematic comparison of six distinct computational methods for detecting viral compartmentalization.
• Utilized three datasets: sequences from the central nervous system (CNS) of 45 patients, female genital tract sequences from 18 patients, and simulated sequences.

Main Results:

• Significant discrepancies were observed between the conclusions drawn by different compartmentalization detection methods.
• Phylogenetic tree-based methods demonstrated higher sensitivity in detecting compartmentalization compared to methods relying solely on pairwise genetic distances.
• Phylogenetic tree uncertainty, particularly with short, low-diversity, or recombinant sequences, can impact the reliability of tree-based approaches.

Conclusions:

• No single method reliably detects HIV compartmentalization; different approaches frequently yield contradictory results.
• Tree-based methods are generally more sensitive but require adjustments to account for phylogenetic uncertainty.
• A combination of multiple analytical approaches is recommended for robust inference of compartmentalized HIV population structures.