Skip to main content

Table 1 Performance comparison between network-based ranking algorithms on different functional similarity protein complex networks

From: A novel method for identifying disease associated protein complexes based on functional similarity protein complex networks

# Protein complex network NBH RWR PRINCE
1 SharedProteinComNet 0.99665 (±0.00988) 0.99134 (±0.01663) 0.99174 (±0.01633)
(P-value = 4.37 × 10−14) (P-value = 1.7 × 10−12)
2 SharedGOTermComNet 0.95869 (±0.07599) 0.90671 (±0.10804) 0.94305 (±0.08432)
(P-value = 1.28 × 10−27) (P-value = 8.2 × 10−4)
3 SharedPPIComNet 0.97704 (±0.05926) 0.92646 (±0.12794) 0.96257 (±0.08166)
(P-value = 1.28 × 10−27) (P-value = 1.51 × 10−4)
  1. –P-values represent the statistical significance of performance comparison between NBH and RWR/PRINCE algorithms. They were calculated using two-sample t-Test for mean assuming unequal variances.
  2. –Data in each cell represent mean (±standard deviation).
  3. –Performance of NBH is statistically significantly better than that of RWR and PRINCE for all three functional similarity protein complex networks. Data row #1, #2 and #3 are detail performance comparison of the three algorithms on SharedProteinComNet, SharedGOTermComNet and SharedPPIComNet networks corresponding to Figure 2(a), (b) and (c), respectively.