### Abstract

Language | English |
---|---|

Title of host publication | Pattern Recognition in Bioinformatics : 6th IAPR International Conference, PRIB 2011, Delft, The Netherlands, 2-4 November 2011 |

Editors | M. Loog, L. Wessels |

Place of Publication | Berlin [etc.] |

Pages | 97-108 |

DOIs | |

Publication status | Published - 2011 |

### Publication series

Name | Lecture notes in computer science |
---|---|

Publisher | Springer |

No. | 7036 |

### Fingerprint

### Cite this

*Pattern Recognition in Bioinformatics : 6th IAPR International Conference, PRIB 2011, Delft, The Netherlands, 2-4 November 2011*(pp. 97-108). (Lecture notes in computer science; No. 7036). Berlin [etc.]. https://doi.org/10.1007/978-3-642-24855-9_9

}

*Pattern Recognition in Bioinformatics : 6th IAPR International Conference, PRIB 2011, Delft, The Netherlands, 2-4 November 2011.*Lecture notes in computer science, no. 7036, Berlin [etc.], pp. 97-108. https://doi.org/10.1007/978-3-642-24855-9_9

**Metabolic Pathway Inference from Time Series Data: A Non Iterative Approach.** / Astola, L.J.; Groenenboom, M.A.C.; Gomez Roldan, M.V.; van Eeuwijk, F.A.; Hall, R.D.; Bovy, A.G.; Molenaar, J.

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

TY - CHAP

T1 - Metabolic Pathway Inference from Time Series Data: A Non Iterative Approach

AU - Astola, L.J.

AU - Groenenboom, M.A.C.

AU - Gomez Roldan, M.V.

AU - van Eeuwijk, F.A.

AU - Hall, R.D.

AU - Bovy, A.G.

AU - Molenaar, J.

PY - 2011

Y1 - 2011

N2 - In this article, we present a very fast and easy to implement method for reconstruction of metabolic pathways based on time series data. To model the metabolic reactions, we use the well-established setting of ordinary differential equations. In the present article we consider a network leading to the accumulation of quercetin-glycosides in tomato (Solanum lycopersicum). Quercetin belongs to a group of plant secondary metabolites, generally referred to as flavonoids, which are extensively being studied for their variety of important functions in plants as well as for their potentially health-promoting effects on human. We use time series measurements of metabolite concentrations of quercetin derivatives. In the present setting, the observed concentrations are the variables and the reaction rates are the unknown parameters. A standard method is to solve the parameters by reverse engineering, where the ordinary differential equations (ODE) are solved repeatedly, resulting in impractical computation times. We use an alternative method that estimates the parameters by least squares minimization, and which is, in the order of hundred times faster than the iterative method. Our reconstruction method can incorporate an arbitrary a priori known network structure as well as positivity constraints on the reaction rates. In this way we can avoid over-fitting, which is another often encountered problem in network reconstruction, and thus obtain better estimates for the parameters. We test the presented method by reconstructing artificial networks and compare it with the more conventional method in terms of residuals between the observed and fitted concentrations, computing times and the proportion of correctly identified edges in the network. Finally we exploit this fast method to statistically infer the kinetic constants in the flavonoid pathway. We remark that the method as such is not limited to metabolic network reconstructions, but can be used with any type of time-series data that is modeled in terms of linear ODE’s.

AB - In this article, we present a very fast and easy to implement method for reconstruction of metabolic pathways based on time series data. To model the metabolic reactions, we use the well-established setting of ordinary differential equations. In the present article we consider a network leading to the accumulation of quercetin-glycosides in tomato (Solanum lycopersicum). Quercetin belongs to a group of plant secondary metabolites, generally referred to as flavonoids, which are extensively being studied for their variety of important functions in plants as well as for their potentially health-promoting effects on human. We use time series measurements of metabolite concentrations of quercetin derivatives. In the present setting, the observed concentrations are the variables and the reaction rates are the unknown parameters. A standard method is to solve the parameters by reverse engineering, where the ordinary differential equations (ODE) are solved repeatedly, resulting in impractical computation times. We use an alternative method that estimates the parameters by least squares minimization, and which is, in the order of hundred times faster than the iterative method. Our reconstruction method can incorporate an arbitrary a priori known network structure as well as positivity constraints on the reaction rates. In this way we can avoid over-fitting, which is another often encountered problem in network reconstruction, and thus obtain better estimates for the parameters. We test the presented method by reconstructing artificial networks and compare it with the more conventional method in terms of residuals between the observed and fitted concentrations, computing times and the proportion of correctly identified edges in the network. Finally we exploit this fast method to statistically infer the kinetic constants in the flavonoid pathway. We remark that the method as such is not limited to metabolic network reconstructions, but can be used with any type of time-series data that is modeled in terms of linear ODE’s.

U2 - 10.1007/978-3-642-24855-9_9

DO - 10.1007/978-3-642-24855-9_9

M3 - Chapter

SN - 9783642248542

T3 - Lecture notes in computer science

SP - 97

EP - 108

BT - Pattern Recognition in Bioinformatics : 6th IAPR International Conference, PRIB 2011, Delft, The Netherlands, 2-4 November 2011

A2 - Loog, M.

A2 - Wessels, L.

CY - Berlin [etc.]

ER -