Optimization of neural network architecture using genetic programming improves detection and modelling of gene-gene interactions in studies of human diseases

Marylyn Deriggi Ritchie, Bill C. White, Joel S. Parker, Lance W. Hahn, Jason H. Moore

Research output: Contribution to journalArticle

166 Citations (Scopus)

Abstract

Background: Appropriate definition of neural network architecture prior to data analysis is crucial for successful data mining. This can be challenging when the underlying model of the data is unknown. The goal of this study was to determine whether optimizing neural network architecture using genetic programming as a machine learning strategy would improve the ability of neural networks to model and detect nonlinear interactions among genes in studies of common human diseases. Results: Using simulated data, we show that a genetic programming optimized neural network approach is able to model gene-gene interactions as well as a traditional back propagation neural network. Furthermore, the genetic programming optimized neural network is better than the traditional back propagation neural network approach in terms of predictive ability and power to detect gene-gene interactions when non-functional polymorphisms are present. Conclusion: This study suggests that a machine learning strategy for optimizing neural network architecture may be preferable to traditional trial-and-error approaches for the identification and characterization of gene-gene interactions in common, complex human diseases.

Original languageEnglish (US)
Article number28
JournalBMC Bioinformatics
Volume4
DOIs
StatePublished - Jul 7 2003

Fingerprint

Genetic programming
Network Architecture
Network architecture
Genetic Programming
Genes
Neural Networks
Gene
Neural networks
Optimization
Interaction
Modeling
Learning Strategies
Back-propagation Neural Network
Backpropagation
Machine Learning
Learning systems
Neural Networks (Computer)
Data Mining
Trial and error
Nonlinear Interaction

All Science Journal Classification (ASJC) codes

  • Structural Biology
  • Biochemistry
  • Molecular Biology
  • Computer Science Applications
  • Applied Mathematics

Cite this

Ritchie, Marylyn Deriggi ; White, Bill C. ; Parker, Joel S. ; Hahn, Lance W. ; Moore, Jason H. / Optimization of neural network architecture using genetic programming improves detection and modelling of gene-gene interactions in studies of human diseases. In: BMC Bioinformatics. 2003 ; Vol. 4.
@article{6dbfdd0771cc48bf9048521574938a1b,
title = "Optimization of neural network architecture using genetic programming improves detection and modelling of gene-gene interactions in studies of human diseases",
abstract = "Background: Appropriate definition of neural network architecture prior to data analysis is crucial for successful data mining. This can be challenging when the underlying model of the data is unknown. The goal of this study was to determine whether optimizing neural network architecture using genetic programming as a machine learning strategy would improve the ability of neural networks to model and detect nonlinear interactions among genes in studies of common human diseases. Results: Using simulated data, we show that a genetic programming optimized neural network approach is able to model gene-gene interactions as well as a traditional back propagation neural network. Furthermore, the genetic programming optimized neural network is better than the traditional back propagation neural network approach in terms of predictive ability and power to detect gene-gene interactions when non-functional polymorphisms are present. Conclusion: This study suggests that a machine learning strategy for optimizing neural network architecture may be preferable to traditional trial-and-error approaches for the identification and characterization of gene-gene interactions in common, complex human diseases.",
author = "Ritchie, {Marylyn Deriggi} and White, {Bill C.} and Parker, {Joel S.} and Hahn, {Lance W.} and Moore, {Jason H.}",
year = "2003",
month = "7",
day = "7",
doi = "10.1186/1471-2105-4-28",
language = "English (US)",
volume = "4",
journal = "BMC Bioinformatics",
issn = "1471-2105",
publisher = "BioMed Central",

}

Optimization of neural network architecture using genetic programming improves detection and modelling of gene-gene interactions in studies of human diseases. / Ritchie, Marylyn Deriggi; White, Bill C.; Parker, Joel S.; Hahn, Lance W.; Moore, Jason H.

In: BMC Bioinformatics, Vol. 4, 28, 07.07.2003.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Optimization of neural network architecture using genetic programming improves detection and modelling of gene-gene interactions in studies of human diseases

AU - Ritchie, Marylyn Deriggi

AU - White, Bill C.

AU - Parker, Joel S.

AU - Hahn, Lance W.

AU - Moore, Jason H.

PY - 2003/7/7

Y1 - 2003/7/7

N2 - Background: Appropriate definition of neural network architecture prior to data analysis is crucial for successful data mining. This can be challenging when the underlying model of the data is unknown. The goal of this study was to determine whether optimizing neural network architecture using genetic programming as a machine learning strategy would improve the ability of neural networks to model and detect nonlinear interactions among genes in studies of common human diseases. Results: Using simulated data, we show that a genetic programming optimized neural network approach is able to model gene-gene interactions as well as a traditional back propagation neural network. Furthermore, the genetic programming optimized neural network is better than the traditional back propagation neural network approach in terms of predictive ability and power to detect gene-gene interactions when non-functional polymorphisms are present. Conclusion: This study suggests that a machine learning strategy for optimizing neural network architecture may be preferable to traditional trial-and-error approaches for the identification and characterization of gene-gene interactions in common, complex human diseases.

AB - Background: Appropriate definition of neural network architecture prior to data analysis is crucial for successful data mining. This can be challenging when the underlying model of the data is unknown. The goal of this study was to determine whether optimizing neural network architecture using genetic programming as a machine learning strategy would improve the ability of neural networks to model and detect nonlinear interactions among genes in studies of common human diseases. Results: Using simulated data, we show that a genetic programming optimized neural network approach is able to model gene-gene interactions as well as a traditional back propagation neural network. Furthermore, the genetic programming optimized neural network is better than the traditional back propagation neural network approach in terms of predictive ability and power to detect gene-gene interactions when non-functional polymorphisms are present. Conclusion: This study suggests that a machine learning strategy for optimizing neural network architecture may be preferable to traditional trial-and-error approaches for the identification and characterization of gene-gene interactions in common, complex human diseases.

UR - http://www.scopus.com/inward/record.url?scp=0642368712&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0642368712&partnerID=8YFLogxK

U2 - 10.1186/1471-2105-4-28

DO - 10.1186/1471-2105-4-28

M3 - Article

VL - 4

JO - BMC Bioinformatics

JF - BMC Bioinformatics

SN - 1471-2105

M1 - 28

ER -