Choosing the right program for the identification of HIV-1 transmission networks from nucleotide sequences sampled from different populations.

HIV-TRACE and Cluster Picker are some of the most widely used programs for identifying HIV-1 transmission networks from nucleotide sequences. However, choosing between these tools is subjective and often a matter of personal preference. Because these softwares use different algorithms to detect HIV-1 transmission networks, their optimal use is better suited with different sequence datasets and under different scenarios. The performance of these tools has previously been evaluated across a range of genetic distance (GD) thresholds without an assessment of the differences in the structure of networks identified. Here, we tested both programs on the same HIV-1 pol sequence dataset (n=2,017) from three Ugandan populations to examine their performance across different risk groups and evaluate the structure of networks identified. HIV-TRACE that uses a single-linkage algorithm identified more nodes in the same networks that were connected by sparse links than Cluster Picker. This suggests that the choice of the program used for identifying networks should depend on the study aims, the characteristics of the population being investigated, dynamics of the epidemic, sampling design and the nature of research questions being addressed for optimum results. HIV-TRACE could be more applicable with larger datasets where the aim is to identify larger clusters that represent distinct transmission chains and in more diverse populations where infection has occurred over a period of time. In contrast, Cluster Picker is applicable in situations where more closely connected clusters are expected in the studied populations.