Fig 1.
Prototypic network structures.
A. Erdős-Rényi network (ER), B. Barabási-Albert network (BA), C. Watts-Strogatz network (WS). To illustrate the typical configurations, all networks have the same size (50 nodes) and they are characterized by the same mean degree (6). The rewiring probability is 0.01 in WS.
Table 1.
Summary statistics computed on time scaled trees (transmission trees and virus genealogies).
Fig 2.
Relationship among transmission history, transmission tree and virus genealogy.
For a given transmission history between hosts (A), we can construct a binary representation, i.e. the transmission history (B). The lower panels (C) show 4 possible virus genealogies of this transmission history invoking our within-host population model.
Fig 3.
Box plots of tree statistics on transmission trees (red) and reconstructed virus genealogies (light blue).
Tree height (A), number of cherries per taxa (B), mean internal/external branch lengths ratio (C), Colless Index (D), Sackin’s Index (E), mean branch length (F). The boxes correspond to the first and third quartiles. The upper/lower whisker extends from the third/first quartile to the highest/lowest value which is within 1.5 IQR from the box, where IQR is the inter-quartile range.
Fig 4.
Box plots of tree statistics on virus genealogies under varying infectivity profiles.
Tree height (A), number of cherries per taxa (B), Sackin’s index (C), mean internal/external branch lengths ratio (D). Box plots limits are as in Fig 3.
Fig 5.
Box plots of tree statistics on virus genealogies for different assumptions on individual heterogeneity.
Tree height (A), number of cherries per taxa (B), Sackin’s index (C), mean internal/external branch lengths ratio (D). Box plots limits are as in Fig 3.
Fig 6.
Distance based and topological tree statistics on virus genealogies as epidemic progresses on a network of size 1000.
Mean branch length (MBL) as function of the number of infected individuals (A) and as function of the number of taxa (sampled infected individuals) (B) for simulated outbreaks on networks of size 1000 as epidemics progress. Note that there is a time interval between infections and diagnoses (which correspond to removal/sampling times). Sackin’s index (C) and number of cherries per taxa (D) as function of the number of taxa in networks of size 1000. The envelopes represent 95% confidence intervals around the medians. The curves are obtained using local regression (LOESS). WS (red), ER (green), BA (blue).
Fig 7.
Mean branch length as function of the number of taxa varying sampling fraction.
Mean branch length as function of sampled hosts, with varying sampling fraction (p = 1-0.25). The envelopes represent 95% confidence intervals around the medians. The curves are obtained using local regression (LOESS). Note that for smaller sampling fractions the envelopes include fewer taxa. WS (red), ER (green), BA (blue).
Table 2.
Parameter estimation for one epidemic spread on an ER network.
Fig 8.
Time-scaled HIV-1 phylogenies from the Swedish epidemic among IDU.
A. The genealogy from a rapid CRF01 outbreak, and B the genealogy from a slower spreading subtype B epidemics. Trees were inferred by a Bayesian skyline coalescent model using BEAST 1.8 [49].
Table 3.
Network type posterior probability for the two Swedish outbreaks.