FIG. 2.

Comparison of various hierarchies for PI3Kc-like domains (cd00142). (A) Comparison between the Conserved Domain Database manually curated hierarchy (56,085 nats; 22 nodes) and a hierarchy generated de novo (56,514 nats; 20 nodes). Nodes are colored (as indicated) depending on whether they correspond (i.e., share ≥90% of their assigned sequences), roughly correspond (share >50%), or fail to correspond. The orange-highlighted line indicates that assignment of node cd05172 as a child of cd05164 lacks statistical support (i.e., the subLLR is negative). (B) Histogram of sequence diversity (SqDv) for the hierarchies shown in (A). (C) Scatter plot of SqDv versus the number of phyla for the curated and de novo hierarchies. RSD, relative standard deviation.