RETRACTED ARTICLE: Detangling PPI networks to uncover functionally meaningful clusters

BMC Systems Biology

Table 3 The performance of Louvain versus Louvain plus DSD at different edge removal thresholds; the results of Louvain are median values from running the algorithm over 10 random permutations of the nodes. We discard clusters of size < 3 and prevent combining clusters when the resulting cluster would have size > 100

Method	Enriched Clusters	# NEC	% NEC	# NEC S	% NEC S
PPI	78.0/382.0 (20.42%)	1543.5	25.31%	634.5	10.41%
4.0	130.0/192.5 (67.53%)	1138.0	18.67%	1007.0	16.52%
4.5	186.0/305.0 (60.98%)	1915.5	31.42%	1297.5	21.28%
5.0	137.0/352.0 (38.92%)	2283.5	37.46%	1017.5	16.69%
5.5	53.5/227.5 (23.52%)	1987.0	32.60%	462.5	7.59%
6.0	40.5/180.5 (22.44%)	1702.5	27.93%	317.5	5.21%

NEC= “Nodes in Enriched Clusters”. We calculate %NEC in two settings: %NEC is enrichment in the GO hierarchy with terms above the fifth level filtered out, and %NEC S uses the same filtered GO hierarchy, but then only gives a node credit if there is a match between one of the node’s labels and one of the terms for which there is GO enrichment for the cluster. At every DSD threshold we tested except 4, the percentage of nodes in enriched clusters is better than Louvain run alone. The S statistic is better at DSD thresholds between 4 and 5, and best at a DSD threshold of 4.5
Bolded values represent the best values achieved for the %NEC and %NEC S statistics comparing the PPI network and different DSD detangling thresholds

ISSN: 1752-0509