Skip to main content

Table 3 Performance summary statistic (AUC and F-score) and percentage consistent GO terms.

From: Predicting functional associations from metabolism using bi-partite network algorithms

Scores Performance on 170 SL pairs (full network) Performance on 69 SL pairs(reduced network) Genes in full network, 283 GO terms (152-BP; 85-MF; 46-CC) Genes in reduced network, 203 GO terms (110-BP; 60-MF; 33-CC)
  AUC
(ROC)
F-Score
(PR)
AUC
(ROC)
F-Score
(PR)
Total BP MF CC Total BP MF CC
Compartments
Shared Mets 0.69 0.16 0.75 0.26 61 52 69 78 50 41 62 61
Shared Mets
(low degree)
0.67 0.15 0.76 0.27 65 64 61 74 55 58 45 61
Hypergeometric
p-value
0.69 0.21 0.76 0.43 72 64 79 85 63 55 75 67
Bayes Score 0.70 0.21 0.76 0.43 72 64 79 83 63 55 75 67
Poisson Score 0.70 0.20 0.76 0.40 72 64 79 85 60 52 73 64
Poisson Score (met. degree) 0.70 0.21 0.79 0.47 81 73 94 83 74 65 90 73
Graph diffusion kernel score 0.74 0.23 0.79 0.39 94 90 99 98 92 87 97 100
Graph diffusion kernel score (weighted edges) 0.72 0.22 0.79 0.39 93 88 99 98 92 87 97 100
Flux Correlation - - 0.87 0.49 - - - - 64 66 53 73
Poisson Score (met. associations, deg.) 0.70 0.21 0.79 0.47 81 73 93 83 74 65 92 70
Scaled epistasis - - 0.54 0.08 - - - - - - - -
No Compartments
Shared mets. 0.68 0.09 0.73 0.22 55 47 69 54 49 38 67 52
Shared mets.
(low deg. mets.)
0.66 0.18 0.73 0.26 71 72 76 54 64 67 62 58
Hypergeometric
P-value
0.69 0.13 0.76 0.39 66 59 78 65 57 45 73 67
Bayes Score 0.69 0.13 0.76 0.39 66 59 78 65 57 45 73 67
Poisson Score 0.69 0.12 0.75 0.33 64 57 76 65 56 44 73 64
Poisson Score
(met. degree)
0.72 0.31 0.80 0.49 78 72 93 72 75 65 92 76
Graph diffusion
kernel score
0.72 0.26 0.79 0.38 94 91 99 93 91 85 98 94
Graph diffusion kernel score
(weighted edges)
0.70 0.25 0.79 0.4 93 90 99 93 89 86 93 88
Poisson score(met. associations, deg.) 0.72 0.30 0.79 0.48 80 74 93 72 75 65 90 79
  1. The best performance in each category is highlighted in bold.