CASCADE: a novel quasi all paths-based network analysis algorithm for clustering biological interactions

BMC Bioinformatics

Table 3 Comparison of CASCADE to competing clustering methods for 3 biological network data sets (Yeast DNA damage response network, Rapamycin gene modules network, and Rich medium gene modules network).

Dataset	Method	Number	Size	Discard(%)	Function (-log p)
DNA damage response network	CASCADE	6	15.7	5.0	2.28
	STM	6	16.0	5.2	2.28
	Quasi clique	3	7.0	88.5	0.87
	Samanta	6	6.7	58.3	1.79
	Minimum cut	7	13.1	4.2	1.18
	Betweenness cut	10	8.8	8.3	2.22
	MCL	3	9.3	70.8	2.37
	Chen	7	13.7	0.0	2.66
	Rives	5	18.4	4.1	1.61
	SPC	3	20.3	36.5	2.33
Rapamycin gene modules network	CASCADE	4	11.8	6.0	2.90
	STM	4	12.5	0.0	2.57
	Quasi clique	13	8.2	0.0	2.17
	Samanta	7	4.9	32.0	1.57
	Minimum cut	8	5.9	6.0	1.82
	Betweenness cut	5	8.0	20.0	2.03
	MCL	6	7.7	8.0	5.48
	Chen	5	10.0	0.0	2.01
	Rives	4	11.0	12.0	1.49
	SPC	3	15.3	8.0	1.47
Rich medium gene modules network	CASCADE	4	27.8	0.0	10.5
	STM	5	22.4	0.0	8.21
	Quasi clique	5	22.8	0.0	7.81
	Samanta	12	5.3	43.2	4.79
	Minimum cut	10	11.1	0.0	4.41
	Betweenness cut	8	13.9	0.0	6.38
	MCL	23	4.0	4.5	7.29
	Chen	8	13.9	0.0	6.13
	Rives	5	22.2	0.0	5.77
	SPC	5	20.6	7.2	6.80

The Number column indicates the number of clusters identified by each method, the Size column indicates the average number of molecular components in each cluster; the Discard(%) indicates the percentage of molecular components not assigned to any cluster. The average -log p values of all detected clusters for biological function are shown. Comparisons are performed on the clusters with 5 or more molecular components for the first data sets(DNA damage response network), and on the clusters with 3 or more molecular components for the last 2 network data sets(Rapamycin gene modules networks, Rich medium gene modules network). The Maximal clique method is not included in the Table because none of the identified clusters have 3 or more members.

ISSN: 1471-2105