Protein domain organisation: adding order

BMC Bioinformatics

Table 1 Global properties of the domain architecture network.

Property	Observed	Random (mean)	Stdev	Deviation from observed
Directed
Mean Degree	3.59	3.59	0	0
Mean Clustering Coefficient	0.14	0.05	0.004	+22.5SD
Characteristic Path Length	3.87	3.83	0.02	+2SD
Network density (% connected)	57.7	60.6	0.58	-5SD
Undirected
Mean Degree	6.47	7.00	0	0
Mean Clustering Coefficient	0.23	0.08	0.007	+20.7SD
Characteristic Path Length	3.40	3.29	0.02	+5.5SD
Network density (% connected)	93.02	94.8	0.64	-2.8SD

The graph represents 6183 distinct domain architectures (from 262,481 proteins in 192 genomes) including 1224 different domain superfamilies. The mean values and standard deviations of each property are for 1000 random graphs, each generated to have the same in- and out-degree as the domain organisation graph (Figure 2). The characteristic paths lengths for directed and undirected graphs are numerically quite similar. However, these numbers cannot be directly compared because the proportion of nodes that are reachable (and hence included in the path length calculation) is much higher in the undirected graph (61% for the directed compared to 95% for the undirected).

ISSN: 1471-2105