Advanced search
1 file | 254.02 KB

Network inference by integrating biclustering and feature selection

Robrecht Cannoodt (UGent) , Joeri Ruyssinck (UGent) , Katleen De Preter (UGent) , Tom Dhaene (UGent) and Yvan Saeys (UGent)
Author
Organization
Abstract
In order to develop better therapies to combat specific abnormalities present in the gene regulatory network (GRN) of cancer patients, it is crucial to gain a better understanding of regulatory networks in complex biological systems. An important class of methods in systems biology are network inference (NI) methods, which aim to reconstruct a GRN from high-throughput data (e.g. microarrays or next-generation sequencing). GENIE3 is a state-of-the-art method which employs feature selection to identify the best subset of regulators for each gene. While this method is amongst the best performing NI methods, it fails to take into account expected topological properties of a GRN: a GRN consists of modules, each of which consists of genes coregulated by a common set of regulators. We present BiGENIE, a method which takes the modular topology of a GRN into account. By firstly inferring modules – groups of genes coregulated by a common regulator – using several biclustering methods, the overall topology of the network is reconstructed. Subsequently, the regulator genes for each of the modules is inferred using GENIE3.
Keywords
network inference, gene regulatory network, machine learning

Downloads

  • BBC2013 robrecht cannoodt.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 254.02 KB

Citation

Please use this url to cite or link to this publication:

Chicago
Cannoodt, Robrecht, Joeri Ruyssinck, Katleen De Preter, Tom Dhaene, and Yvan Saeys. 2013. “Network Inference by Integrating Biclustering and Feature Selection.” In Proceedings of BBC2013, 33–33.
APA
Cannoodt, R., Ruyssinck, J., De Preter, K., Dhaene, T., & Saeys, Y. (2013). Network inference by integrating biclustering and feature selection. Proceedings of BBC2013 (pp. 33–33). Presented at the BeNeLux Bioinformatics Conference 2013.
Vancouver
1.
Cannoodt R, Ruyssinck J, De Preter K, Dhaene T, Saeys Y. Network inference by integrating biclustering and feature selection. Proceedings of BBC2013. 2013. p. 33–33.
MLA
Cannoodt, Robrecht et al. “Network Inference by Integrating Biclustering and Feature Selection.” Proceedings of BBC2013. 2013. 33–33. Print.
@inproceedings{4252784,
  abstract     = {In order to develop better therapies to combat specific abnormalities present in the gene regulatory network (GRN) of cancer patients, it is crucial to gain a better understanding of regulatory networks in complex biological systems. An important class of methods in systems biology are network inference (NI) methods, which aim to reconstruct a GRN from high-throughput data (e.g. microarrays or next-generation sequencing). GENIE3 is a state-of-the-art method which employs feature selection to identify the best subset of regulators for each gene. While this method is amongst the best performing NI methods, it fails to take into account expected topological properties of a GRN: a GRN consists of modules, each of which consists of genes coregulated by a common set of regulators. We present BiGENIE, a method which takes the modular topology of a GRN into account. By firstly inferring modules -- groups of genes coregulated by a common regulator -- using several biclustering methods, the overall topology of the network is reconstructed. Subsequently, the regulator genes for each of the modules is inferred using GENIE3.},
  author       = {Cannoodt, Robrecht and Ruyssinck, Joeri and De Preter, Katleen and Dhaene, Tom and Saeys, Yvan},
  booktitle    = {Proceedings of BBC2013},
  language     = {eng},
  location     = {Brussels, Belgium},
  pages        = {33--33},
  title        = {Network inference by integrating biclustering and feature selection},
  year         = {2013},
}