Skip to main content

Table 2 The performance of Walktrap versus Walktrap plus DSD at different edge removal thresholds; We discard clusters of size < 3

From: RETRACTED ARTICLE: Detangling PPI networks to uncover functionally meaningful clusters

Method Enriched Clusters # NEC % NEC # NEC S % NEC S
PPI 8/19 (42.11%) 280.0 4.59% 226.0 3.71%
3.5 63/105 (60.00%) 504.0 8.27% 464.0 7.61%
4.0 128/189 (67.72%) 1108.0 18.18% 919.0 15.08%
4.5 207/311 (66.56%) 1951.0 32.00% 1430.0 23.46%
5.0 153/303 (50.50%) 2476.0 40.62% 1531.0 25.11%
5.5 70/164 (42.68%) 2418.0 39.67% 1269.0 20.82%
6.0 43/88 (48.86%) 1398.0 22.93% 837.0 13.73%
  1. 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. Walktrap run alone produces a very small number of clusters; because of this only the S statistic is meaningful to compare the DSD versions against unmodified Walktrap. Walktrap with DSD at thresholds between 4.5 and 6 trade a larger number of smaller clusters for a lower percentage of nodes in enriched clusters
  2. Bolded values represent the best values achieved for the %NEC and %NEC S statistics comparing the PPI network and different DSD detangling thresholds