The phylogenetic trees of VP1 and RdRp-encoding nucleotide sequences were inferred using IQ-TREE v1.6.12 [47 (link)] with 10,000 pseudo-replicates [48 (link)], incorporating the best-fit model of nucleotide substitution (VP1: TIM2 + F + R10, RdRp: GTR + F + R10) [49 (link)], and rooted by a midpoint. Trees were visualized with ggtree R-package [50 (link)].
Maximum clade credibility (MCC) trees for sequences with available collection dates were inferred for GI (N = 71 sequences) and GII (N = 915 sequences) using BEAST v.1.10.4 [51 (link)]. The best-fit partitioning scheme (GI: (1 + 2)(3), GII: (1,2,3)) and substitution models (GTR + I + G + X) for Bayesian analysis were chosen according to the Bayesian Information Criterion using the PartitionFinder 2 program [52 (link)]. For each genogroup, marginal likelihoods were calculated for combinations of coalescent tree priors (coalescent constant size, coalescent exponential growth) and molecular clock models (strict, relaxed log-normal) using the path sampling/stepping stone procedure implemented in BEAST v1.10.4 [53 (link)]. Then, different model settings were compared using the Bayes factor (BF) test. The combination of coalescent constant prior and relaxed lognormal molecular clock was strongly favored (log BF > 10) for both genogroups. The MCMC chains were run for 50 and 800 million steps with sampling every 5000 and 10,000 steps for GI и GII, respectively. The convergence of Markov chain Monte Carlo (MCMC) was inspected using Tracer v1.7 [54 (link)]. The maximum clade credibility (MCC) tree was annotated with TreeAnnotator v1.10.4 using 10% burn-in.
Maximum clade credibility (MCC) trees for sequences with available collection dates were inferred for GI (N = 71 sequences) and GII (N = 915 sequences) using BEAST v.1.10.4 [51 (link)]. The best-fit partitioning scheme (GI: (1 + 2)(3), GII: (1,2,3)) and substitution models (GTR + I + G + X) for Bayesian analysis were chosen according to the Bayesian Information Criterion using the PartitionFinder 2 program [52 (link)]. For each genogroup, marginal likelihoods were calculated for combinations of coalescent tree priors (coalescent constant size, coalescent exponential growth) and molecular clock models (strict, relaxed log-normal) using the path sampling/stepping stone procedure implemented in BEAST v1.10.4 [53 (link)]. Then, different model settings were compared using the Bayes factor (BF) test. The combination of coalescent constant prior and relaxed lognormal molecular clock was strongly favored (log BF > 10) for both genogroups. The MCMC chains were run for 50 and 800 million steps with sampling every 5000 and 10,000 steps for GI и GII, respectively. The convergence of Markov chain Monte Carlo (MCMC) was inspected using Tracer v1.7 [54 (link)]. The maximum clade credibility (MCC) tree was annotated with TreeAnnotator v1.10.4 using 10% burn-in.
Free full text: Click here
