Figure 2

The frequencies of the three different PSZ thresholds are found on 1571 protein families (all COGs with more than 25 sequences). The normalized frequencies are represented as kernel density estimates. All optimal ranges used consist of two maximum densities from -1.0 to -3.0 (PSZ = -3.04 and -1.82), from -1.0 to -2.5 (PSZ = -2.60 and -1.74), and from -1.0 to -2.0 (PSZ = -2.01 and -1.58). Kernel densities from fifteen manually set thresholds that have been structurally verified also contain two maximum densities (PSZ = -2.12 and -1.53). Of the three ranges tested, the {-1.0: -2.0} gap is the only one that is statistically similar to the structurally verified results. The coloring of the three ranges is the same as in Figure 2C.