Figure 4.

Definition of Temporal Classes of VACV Gene Expression

(A) Number of temporal classes of VACV gene expression. The k-means approach was used with 1–15 classes to cluster viral proteins, and the summed distance of each protein from its cluster centroid was calculated. Although this summed distance necessarily becomes smaller as more clusters are added, the rate of decline decreases with each added group, eventually settling at a fairly constant rate of decline that reflects overfitting; clusters added prior to this point reflect the underlying structure in the temporal protein data, whereas clusters subsequently added through overfitting are not informative. The point of inflexion fell between four and six classes, suggesting that there are at least four distinct temporal protein profiles of viral protein expression.

(B) Class centroid profiles.

(C) Number of viral proteins per class, and the number of proteins in each class whose expression was reduced >1.5-fold by incubation with AraC (treated and untreated samples both assessed at 6 h of infection; see schematic in Figure 1A and results in Figure S3).

(D) Temporal profiles of proteins in each k-means class were subjected to hierarchical clustering by Euclidian distance.

(E) Temporal profiles of representative proteins from each cluster. Data are represented as mean ± SEM (n = 3).

(F) Comparison of viral protein and transcript classes.

(G) Functional analysis of viral proteins, based on information from Yang et al. (2010) and other references detailed in Table S5.