## Finding Optimal Gene Networks Using Biological Constraints (2003)

Venue: | Genome Informatics |

Citations: | 12 - 3 self |

### BibTeX

@ARTICLE{Ott03findingoptimal,

author = {Sascha Ott and Satoru Miyano},

title = {Finding Optimal Gene Networks Using Biological Constraints},

journal = {Genome Informatics},

year = {2003},

volume = {14},

pages = {2003}

}

### OpenURL

### Abstract

The accurate estimation of gene networks from gene expression measurements is a major challenge in the field of Bioinformatics. Since the problem of estimating gene networks is NP-hard and exhibits a search space of super-exponential size, researchers are using heuristic algorithms for this task. However, little can be said about the accuracy of heuristic estimations. In order to overcome this problem, we present a general approach to reduce the search space to a biologically meaningful subspace and to find optimal solutions within the subspace in linear time. We show the e#ectiveness of this approach in application to yeast and Bacillus subtilis data.

### Citations

1127 | A Bayesian method for the induction of probabilistic networks from data. Machine Learning 9:309–347
- Cooper, Herskovits
- 1992
(Show Context)
Citation Context ... decomposed in conditional probability distributions using a directed acyclic graph, which we will call network. Researchers have proposed a number of score functions for the selection of the network =-=[3, 6, 11]-=-, given gene expression measurement data. Work on new score functions exploiting previous knowledge is on-going [13, 22]. Having selected a score function, the task of estimating a gene network is to ... |

872 |
Comprehensive identification of cell cycle-regulated genes of the yeast S. cerevisiae by microarray hybridization
- Spellman, Sherlock, et al.
- 1998
(Show Context)
Citation Context ...order to find a biologically meaningful subspace. Here, we apply Algorithm 3 to RNA microarray data for Saccharomyces cerevisiae obtained from time-course experiments using synchronised cell cultures =-=[21]-=-. This data set can be expected to contain some information about the gene network that works during the cell cycle of yeast. 1 1 The actual probability is lower than , because we count breaking ties ... |

794 | Using Bayesian Network to Analyze Expression Data
- Friedman, Linial, et al.
- 2000
(Show Context)
Citation Context ...h in recent years. If successful, this is expected to yield a wealth of knowledge with applications in a number of fields [23]. A frequently used approach to model gene networks are Bayesian networks =-=[6, 7, 8, 11, 12, 14, 16, 19]-=-. In Bayesian networks, the behaviour of the network is described by a joint probability distribution for the expression levels of all genes. The joint probability distribution must be decomposed in c... |

404 |
Network motifs in the transcriptional regulation network of Escherichia coli
- Shen-Orr, Milo, et al.
(Show Context)
Citation Context ...onents in the graph containing all candidate interactions can be bound by a constant. Since gene networks are assumed to consist of highly connected blocks, which are sparsely connected to each other =-=[10, 18]-=-, this restriction seems to be satisfiable in many research settings. Therefore, Theorem 2 provides the basis for a general approach of gene network estimation. In this work, we give two examples of a... |

378 |
A genome-wide transcriptional analysis of the mitotic cell cycle
- Cho, Campbell, et al.
- 1998
(Show Context)
Citation Context ...in the network, rows childs. It is known that the three gene pairs cln1/cln2, clb5/clb6, and clb1/clb2 are activated specificly during the G1/S phase, the S phase, resp. the M phase of the cell cycle =-=[2]-=-. By a clustering analysis based on this fact, several genes were predicted to be predominantly activated during one of these phases [13]. We selected a set of 43 genes, divided them into three groups... |

336 |
Frommolecular to modular cell biology
- Hartwell
- 1999
(Show Context)
Citation Context ...onents in the graph containing all candidate interactions can be bound by a constant. Since gene networks are assumed to consist of highly connected blocks, which are sparsely connected to each other =-=[10, 18]-=-, this restriction seems to be satisfiable in many research settings. Therefore, Theorem 2 provides the basis for a general approach of gene network estimation. In this work, we give two examples of a... |

243 | Learning Bayesian networks with local structure
- Friedman, Goldszmidt
- 1998
(Show Context)
Citation Context ...h in recent years. If successful, this is expected to yield a wealth of knowledge with applications in a number of fields [23]. A frequently used approach to model gene networks are Bayesian networks =-=[6, 7, 8, 11, 12, 14, 16, 19]-=-. In Bayesian networks, the behaviour of the network is described by a joint probability distribution for the expression levels of all genes. The joint probability distribution must be decomposed in c... |

166 | Inferring Subnetworks from Perturbed Expression Profiles - Pe'er, Regev, et al. - 2001 |

139 | Using graphical models and genomic expression data to statistically validate models of genetic regulatory networks
- Hartemink, Gifford, et al.
- 2001
(Show Context)
Citation Context ...h in recent years. If successful, this is expected to yield a wealth of knowledge with applications in a number of fields [23]. A frequently used approach to model gene networks are Bayesian networks =-=[6, 7, 8, 11, 12, 14, 16, 19]-=-. In Bayesian networks, the behaviour of the network is described by a joint probability distribution for the expression levels of all genes. The joint probability distribution must be decomposed in c... |

103 | Combining location and expression data for principled discovery of genetic regulatory network models
- Hartemink, Gifford, et al.
- 2002
(Show Context)
Citation Context ...ver, facing the NP-hard gene network estimation problem [1] and a search space of superexponential size [17], researchers apply heuristic algorithms like greedy algorithms [11] or simulated annealing =-=[9]-=- in order to estimate gene networks. Since heuristic algorithms do not provide any assurance on their accuracy, it remains unclear to which extent the network estimation will be biased by the computat... |

98 | Miyano: Estimation of genetic networks and functional structures between genes by using Bayesian networks and nonparametric regression. Pac Symp Biocomput175-86
- Imoto, Goto, et al.
- 2002
(Show Context)
Citation Context |

77 | A Comparison of Genetic Network Models
- Wessels, Someren, et al.
- 2001
(Show Context)
Citation Context ...The elucidation of gene interactions in cells is an important focus of research in recent years. If successful, this is expected to yield a wealth of knowledge with applications in a number of fields =-=[23]-=-. A frequently used approach to model gene networks are Bayesian networks [6, 7, 8, 11, 12, 14, 16, 19]. In Bayesian networks, the behaviour of the network is described by a joint probability distribu... |

61 |
Estimating Gene Networks from Gene Expression Data by Combining Bayesian Network Model with Promoter Element Detection
- Tamada, Kim, et al.
- 2003
(Show Context)
Citation Context ...rs have proposed a number of score functions for the selection of the network [3, 6, 11], given gene expression measurement data. Work on new score functions exploiting previous knowledge is on-going =-=[13, 22]-=-. Having selected a score function, the task of estimating a gene network is to find a network with minimal score. However, facing the NP-hard gene network estimation problem [1] and a search space of... |

58 |
Two yeast forkhead genes regulate the cell cycle and pseudohyphal growth
- Zhu, Spellman, et al.
- 2000
(Show Context)
Citation Context ... effectively, while the computation time does not impose practical limitations for the number of genes. 2 Interestingly, swi5 is predicted to be regulated by fkh2 which is a known regulatory relation =-=[24]-=-.s132 Ott and Miyano Table 4: Molecular functions of genes with at most 2 predicted interactions. Gene Number of Interactions Molecular Function sco2 1 unknown sto1 1 mRNA binding bbp1 2 structural co... |

53 | Modelling regulatory pathways in E. coli from time series expression profiles
- Ong, Glasner, et al.
- 2002
(Show Context)
Citation Context |

38 | Bayesian network and nonparametric heteroscedastic regression for nonlinear modeling of genetic network
- Imoto, Kim, et al.
(Show Context)
Citation Context |

38 | Evaluating Functional Network Inference Using Simulations of Complex Biological Systems. (2002) Accepted by The 10th international conference on Intelligent Systems for Molecular Biology
- Smith, Jarvis, et al.
(Show Context)
Citation Context |

32 |
Open source clustering software. Bioinformatics 2004;20(9):1453–1454
- MJ, Imoto, et al.
(Show Context)
Citation Context ...ithm 2 and Algorithm 3 can be used. 4 Results We have implemented Algorithm 2 and Algorithm 3, making use of an existing implementation of Algorithm 1 [15], and publicly available clustering software =-=[4]-=-. As score function s we chose the BNRC score [11], since it can model non-linear gene interactions and can handle the gene expression data without discretization. 4.1 Application of Algorithm 2 to Ba... |

27 | Miyano S . Finding optimal models for small gene networks
- Imoto
(Show Context)
Citation Context ...iologically relevant subspace of the search space is selected. Then, we find optimal solutions in the selected subspace by repeatedly applying an algorithm that computes small gene networks optimally =-=[15]-=-. The selection of the subspace can be done by applying biological knowledge or, if no previous knowledge is available, by using clustering methods. Our approach allows to make use of biological knowl... |

26 |
Learning Bayesian networks is NP-complete,” Lect. Notes Stat
- Chickering
- 1996
(Show Context)
Citation Context ...edge is on-going [13, 22]. Having selected a score function, the task of estimating a gene network is to find a network with minimal score. However, facing the NP-hard gene network estimation problem =-=[1]-=- and a search space of superexponential size [17], researchers apply heuristic algorithms like greedy algorithms [11] or simulated annealing [9] in order to estimate gene networks. Since heuristic alg... |

23 | Using Protein-protein Interactions for Refining Gene Networks Estimated from Microarray Data by Bayesian Networks
- Nariai, Kim, et al.
- 2004
(Show Context)
Citation Context ...rs have proposed a number of score functions for the selection of the network [3, 6, 11], given gene expression measurement data. Work on new score functions exploiting previous knowledge is on-going =-=[13, 22]-=-. Having selected a score function, the task of estimating a gene network is to find a network with minimal score. However, facing the NP-hard gene network estimation problem [1] and a search space of... |

19 |
Bacillus subtilis and its closest relatives: from genes to cells
- Sonenshein, Hoch, et al.
- 2002
(Show Context)
Citation Context ...orithm 2 to Bacillus subtilis Data We have applied Algorithm 2 to Bacillus subtilis data [5]. We selected the data for 6 sigma factors and 79 operons known to be regulated by the chosen sigma factors =-=[20]-=-. The expression ratios of operonssFinding Optimal Gene Networks Using Biological Constraints 129 were defined as the average of the expression ratios of their respective genes. Therefore, we have 79 ... |

17 |
Counting labeled acyclic digraphs. New directions in the theory of graphs
- Robinson
- 1973
(Show Context)
Citation Context ...re function, the task of estimating a gene network is to find a network with minimal score. However, facing the NP-hard gene network estimation problem [1] and a search space of superexponential size =-=[17]-=-, researchers apply heuristic algorithms like greedy algorithms [11] or simulated annealing [9] in order to estimate gene networks. Since heuristic algorithms do not provide any assurance on their acc... |

1 |
Validation of noisy dynamical system models of gene regulation inferred from time-course gene expression data at arbitrary time intervals
- Hoon, Ott, et al.
- 2003
(Show Context)
Citation Context ...linear gene interactions and can handle the gene expression data without discretization. 4.1 Application of Algorithm 2 to Bacillus subtilis Data We have applied Algorithm 2 to Bacillus subtilis data =-=[5]-=-. We selected the data for 6 sigma factors and 79 operons known to be regulated by the chosen sigma factors [20]. The expression ratios of operonssFinding Optimal Gene Networks Using Biological Constr... |