TY - GEN
T1 - GeNet
T2 - 10th European Conference on Evolutionary Computation, Machine Learning and Data Mining in Bioinformatics, EvoBIO 2012
AU - Vanneschi, Leonardo
AU - Mondini, Matteo
AU - Bertoni, Martino
AU - Ronchi, Alberto
AU - Stefano, Mattia
N1 - Vanneschi, L., Mondini, M., Bertoni, M., Ronchi, A., & Stefano, M. (2012). GeNet: A graph-based genetic programming framework for the reverse engineering of gene regulatory networks. In Evolutionary Computation, Machine Learning and Data Mining in Bioinformatics - 10th European Conference, EvoBIO 2012, Proceedings (pp. 97-109). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7246 LNCS). https://doi.org/10.1007/978-3-642-29066-4_9
PY - 2012
Y1 - 2012
N2 - A standard tree-based genetic programming system, called GRNGen, for the reverse engineering of gene regulatory networks starting from time series datasets, was proposed in EvoBIO 2011. Despite the interesting results obtained on the simple IRMA network, GRNGen has some important limitations. For instance, in order to reconstruct a network with GRNGen, one single regression problem has to be solved by GP for each gene. This entails a clear limitation on the size of the networks that it can reconstruct, and this limitation is crucial, given that real genetic networks generally contain large numbers of genes. In this paper we present a new system, called GeNet, which aims at overcoming the main limitations of GRNGen, by directly evolving entire networks using graph-based genetic programming. We show that GeNet finds results that are comparable, and in some cases even better, than GRNGen on the small IRMA network, but, even more importantly (and contrarily to GRNGen), it can be applied also to larger networks. Last but not least, we show that the time series datasets found in literature do not contain a sufficient amount of information to describe the IRMA network in detail.
AB - A standard tree-based genetic programming system, called GRNGen, for the reverse engineering of gene regulatory networks starting from time series datasets, was proposed in EvoBIO 2011. Despite the interesting results obtained on the simple IRMA network, GRNGen has some important limitations. For instance, in order to reconstruct a network with GRNGen, one single regression problem has to be solved by GP for each gene. This entails a clear limitation on the size of the networks that it can reconstruct, and this limitation is crucial, given that real genetic networks generally contain large numbers of genes. In this paper we present a new system, called GeNet, which aims at overcoming the main limitations of GRNGen, by directly evolving entire networks using graph-based genetic programming. We show that GeNet finds results that are comparable, and in some cases even better, than GRNGen on the small IRMA network, but, even more importantly (and contrarily to GRNGen), it can be applied also to larger networks. Last but not least, we show that the time series datasets found in literature do not contain a sufficient amount of information to describe the IRMA network in detail.
UR - http://www.scopus.com/inward/record.url?scp=84859145988&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-29066-4_9
DO - 10.1007/978-3-642-29066-4_9
M3 - Conference contribution
AN - SCOPUS:84859145988
SN - 9783642290657
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 97
EP - 109
BT - Evolutionary Computation, Machine Learning and Data Mining in Bioinformatics - 10th European Conference, EvoBIO 2012, Proceedings
Y2 - 11 April 2012 through 13 April 2012
ER -