Table 1.
Summary of prediction methods using machine learning and network topological features alone or combined with other features*.
| Organisms | ML algorithms | Network type | Network features | Combined with other features? | Train/test** | References |
|---|---|---|---|---|---|---|
| S. cerevisiae | NN, SVM | PIN, GCN | DC | No | Same | Chen and Xu, 2005 |
| S. cerevisiae | WKNN, SVM, ensemble | PIN | DC | Sequence-related | Same | Saha and Heber, 2006 |
|
S. cerevisiae E. coli |
NB | PIN | DC | Sequence-related | Same | Gustafson et al., 2006 |
| E. coli | C4.5 decision tree | PIN, TRN, MN | DC | No | Same | Silva et al., 2008 |
|
S. cerevisiae E. coli |
SVM | PIN | DC, BC, CC, KL, CCo, EI, CFD | Sequence-related | Same | Hwang et al., 2009 |
| S. cerevisiae | Decision tree-based ensemble for prediction; single C4.5 decision tree for description | PIN, TRN, MN | DC, BC, CC, CCo, identicalness | Related to functional annotation | Same | Acencio and Lemke, 2009 |
| S. cerevisiae | GEP | PIN | DC, BC, CC, SC, EC, IC, NC, PeC, WDC, ION | Related to functional annotation | Same | Zhong et al., 2013 |
| P. aeruginosa, E. coli, S. typhinurium | SVM | MN | RUP, PUP, ND, APL, LSP, NS, NP, NNR, NNNR, CCV, DIR, CP, LS, NDR, NDC, NDRD, NDCD, NDCR, NDCC, NDCRD, NDCCD, BC, CC, EC, eccentricity centrality | No | Different | Plaimas et al., 2010 |
| E. coli, P. aeruginosa | Ensemble | GCN | DC, BC | Sequence and gene expression-related | Different | Deng et al., 2011 |
| E. coli, S. cerevisiae, S. sanguinis, S. pombe | FWM (NB, logistic regression, genetic algorithm) | PIN | DC, CC, BC, CCo | Sequence and gene expression-related | Different | Cheng et al., 2013 |
| E. coli, S. enterica, H. influenzae, V. cholerae, P. aeruginosa, Acinetobacter, F. tularensis, H. pylori, C. jejuni, C. crescentus, B. subtilis, S. aureus, S. pneumoniae, S. sanguinis, M. genitalium, M. pulmonis, M. tuberculosis, B. thetaiotaomicron, P. gingivalis, S. cerevisiae, S. pombe | NB | PIN | DC, CC, BC, CCo | Sequence and gene expression-related | Different | Cheng et al., 2014 |
| N. crassa, A. fumigatus | Ensemble | GCN | DC, BC | Sequence and gene expression-related | Different | Lu et al., 2014 |
Abbreviations: NN, neural network; WKNN, weighted k-nearest-neighbor; SVM, support vector machine; NB, Naive bayes; GEP, gene expression programming; FWM, feature-based weighted Naïve Bayes model; PIN, protein-protein interaction network; GCN, gene co-expression network; TRN, transcriptional regulatory network; MN, metabolic network; DC, degree centrality; BC, betweenness centrality; CC, closeness centrality; KL, clique level; CCo, clustering coefficient; EI, essentiality index; CFD, common function degree; SC, subgraph centrality; EC, eigenvector centrality; IC, information centrality; NC, edge-clustering coefficient centrality; WDC, weighted degree centrality; RUP, reachable/unreachable products; PUP, percentage of unreachable products; ND, number of deviations; APL, average path length; LSP, length of the shortest path; NS, number of substrates; NP, number of products; NNR, number of neighboring reactions; NNNR, number of neighboring reactions; CCV, clustering coefficient value; DIR, directionality of a reaction; CP, choke point; LS, load score; NDR, number of damaged reactions; NDC, number of damaged compounds; NDRD, number of damaged reactions having no deviations; NDCR, number of damaged choke; NDCC, number of damaged choke point compounds; NDCRD, number of damaged choke point reactions having no deviations; NDCCD, number of damaged choke point compounds having no deviations.
Same, the sources of training and testing data sets are from same organisms; Different, the sources of training and testing data sets are from different organisms.