Skip to main content

Table 12 The F-score performances of SLPC on denoising PPI networks with different denoising thresholds.

From: Protein complex detection in PPI networks based on data integration and supervised learning method

Threshold DIP Krogan Gavin Avg.Δ
  #Den. #Den.Ext. #Den. #Den.Ext. #Den. #Den.Ext  
0.5 0.5815 0.5889 0.5393 0.5761 0.6789 0.7006 3.76%
0.6 0.5849 0.5912 0.543 0.5854 0.6789 0.7021 4.10%
0.7 0.586 0.5905 0.5418 0.5778 0.6789 0.7012 3.57%
0.8 0.5834 0.5939 0.5414 0.5778 0.6767 0.7001 3.99%
0.9 0.5852 0.5962 0.5456 0.5819 0.6839 0.7057 3.91%
1.0 0.5881 0.596 0.5503 0.5864 0.6855 0.7072 3.69%
1.1 0.5538 0.5785 0.5624 0.5993 0.6627 0.7006 5.58%
1.2 0.5568 0.5776 0.5645 0.5972 0.6634 0.7015 5.09%
1.3 0.5572 0.582 0.5691 0.5984 0.6634 0.7011 5.09%
1.4 0.5537 0.5845 0.565 0.5989 0.6672 0.7065 5.82%
ClusterONE(0.9) 0.4412 0.4241 0.4834 0.4847 0.6418 0.6710 0.31%
Δ(0.9)   40.58%   20.05%   17.42% 26.02%
  1. #Den. denotes the denoising PPI network. #Den.Ext. denotes the denoising extended PPI network. Avg.Δ denotes the average F-score improvement with the different denoising threshold over that on the corresponding denoising networks. Δ(0.9) denotes the improvement of SLPC over ClusterONE with the denoising threshold 0.9.