Influenza viruses undergo continual antigenic evolution allowing mutant viruses to evade host immunity acquired to previous virus strains. from influenza lineages A/H3N2 A/H1N1 B/Victoria and B/Yamagata we determine patterns of antigenic drift across viral lineages showing that A/H3N2 evolves Raf265 derivative faster and in a more punctuated fashion than other influenza lineages. We also show that year-to-year antigenic drift appears to drive incidence patterns within each influenza lineage. This work makes possible substantial future advances in Raf265 derivative investigating the dynamics of influenza and other antigenically-variable pathogens by providing a model that intimately combines molecular and antigenic evolution. DOI: http://dx.doi.org/10.7554/eLife.01914.001 = 0.098) thus supporting our model assumption that the drop in log2 titer is proportional to the Euclidean distance separating viruses and sera on the antigenic map. Additionally we find that the absolute error in predicted titer is nearly constant with time (Pearson correlation = ?0.007). Antigenic locations inferred by the model are well resolved; estimates of antigenic distance between pairs of viruses show relatively little variation across the posterior. We estimate that virus distances have on average a 50% credible interval of ±0.45 antigenic units for A/H3N2 ±0.57 units for A/H1N1 ±0.76 units for B/Vic and ±0.65 units for B/Yam. We find strong correspondence between our results and previous results by Smith et al. (2004) with equivalent models producing globally consistent antigenic maps and other models producing locally consistent maps with a small degree Raf265 derivative of global inconsistency (see ‘Methods’). When implementing the same underlying model differences in the MDS and BMDS approaches reflect greater philosophical differences between maximum-likelihood and Bayesian statistical approaches with the former seeking the single most likely explanation for the data and the latter seeking to fully characterize model uncertainty. Additionally the BMDS method improves flexibility allowing extensions to the basic cartographic model such as the incorporation of virus avidities and evolutionary priors that improve fit and add biological interpretability. Antigenic evolution across influenza lineages Through our analysis we reveal the Raf265 derivative antigenic aswell as evolutionary interactions among infections in influenza A/H3N2 A/H1N1 B/Vic and B/Yam quantifying both antigenic and evolutionary ranges between strains (Body 2 Body 2-supply data 1). More than the period of time of 1968 to 2011 influenza A/H3N2 displays substantially even more antigenic advancement than is certainly exhibited by A/H1N1 Rabbit polyclonal to AGBL2. during the period of 1977 to 2009 or B/Vic and B/Yam during the period of 1986 to 2011. We see prominent antigenic clusters in A/H3N2 and A/H1N1 but much less prominent though still obvious clustering in B/Vic and B/Yam. Antigenic clusters present high hereditary similarity in order that we observe hardly any mutation events resulting in each cluster as opposed to the repeated introduction of clusters. This evaluation makes the destiny of antigenic clusters apparent with two clusters in A/H3N2 (Victoria/75 and Beijing/89) showing up to become evolutionary dead-ends. Labeling of prominent antigenic clusters in Body 2 is supposed as a tough information for orientation rather than as exhaustive catalog of antigenic variant. Figure 2. Antigenic locations of A/H3N2 A/H1N1 B/Yam and B/Vic viruses showing evolutionary relationships between virus samples. HI assays absence sensitivity beyond a particular point in order that for A/H3N2 cross-reactive measurements only exist between strains sampled at most 14 years apart leaving only threshold titers for example ‘<40’ in more temporally distant comparisons. Because of the threshold of sensitivity of the HI assay it is difficult to distinguish a linear trajectory in 2D antigenic space from a slightly curved trajectory (see ‘Materials and Methods’). To solve this problem of identifiability we assumed a poor prior that favors linear movement in the 2D antigenic space (present in models 6 through 9; Table 1) with the slope of the linear relationship and the precision of the relationship incorporated into the Bayesian model (see ‘Materials and methods’). Because of this we interpret map locations locally rather than globally and assess rates of antigenic movement without making strong statements about the larger configuration under which the movement occurs. We find that influenza A/H3N2 evolved along antigenic dimension 1 at an estimated rate Raf265 derivative of 1 Raf265 derivative 1.01 antigenic units per year (Figure 3 Figure 3-source data 1; Table.