| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 3428716 | Virus Research | 2012 | 7 Pages |
Phylogenetic analysis has been extensively used as a tool for the reconstruction of epidemiological relations for research or for forensic purposes. It was our objective to assess the sensitivity of different phylogenetic methods and various phylogenetic programs to reconstruct epidemiological links among HIV-1 infected patients that is the probability to reveal a true transmission relationship. Multiple datasets (90) were prepared consisting of HIV-1 sequences in protease (PR) and partial reverse transcriptase (RT) sampled from patients with documented epidemiological relationship (target population), and from unrelated individuals (control population) belonging to the same HIV-1 subtype as the target population. Each dataset varied regarding the number, the geographic origin and the transmission risk groups of the sequences among the control population. Phylogenetic trees were inferred by neighbor-joining (NJ), maximum likelihood heuristics (hML) and Bayesian methods. All clusters of sequences belonging to the target population were correctly reconstructed by NJ and Bayesian methods receiving high bootstrap and posterior probability (PP) support, respectively. On the other hand, TreePuzzle failed to reconstruct or provide significant support for several clusters; high puzzling step support was associated with the inclusion of control sequences from the same geographic area as the target population. In contrary, all clusters were correctly reconstructed by hML as implemented in PhyML 3.0 receiving high bootstrap support. We report that under the conditions of our study, hML using PhyML, NJ and Bayesian methods were the most sensitive for the reconstruction of epidemiological links mostly from sexually infected individuals.
► Phylogenetic analyses are used for the reconstruction of epidemiological relations. ► We assess the sensitivity of different phylogenetic methods and programs. ► We attempt to reconstruct epidemiological links among HIV-1 infected patients. ► HIV-1 sequences in protease (PR) and partial reverse transcriptase (RT) were used. ► NJ, Bayesian and hML (PhyML 3.0) methods found to be the most sensitive.
