Optimization of neural network architecture using genetic programming improves detection and modeling of gene-gene interactions in studies of human diseases - PubMed (original) (raw)
Optimization of neural network architecture using genetic programming improves detection and modeling of gene-gene interactions in studies of human diseases
Marylyn D Ritchie et al. BMC Bioinformatics. 2003.
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.
Figures
Figure 1
Binary expression tree example of a GP solution. This figure is an example of a possible computer program generated by GP. While the program can take virtually any form, we are using a binary expression tree representation, thus we have shown this type as an example.
Figure 2
GP crossover. This figure shows a crossover event in GP between two binary expression trees. Here, the left sub-tree of parent 1 is swapped with the left sub-tree of parent 2 to create 2 new trees.
Figure 3
GPNN representation of a NN. This figure is an example of one NN optimized by GPNN. The O is the output node, S indicates the activation function, W indicates a weight, and X1-X4 are the NN inputs.
Figure 4
Feed-forward BPNN representation of the GPNN in Figure 3. To generate this NN, each weight in Figure 3 was computed to produce a single value.
Figure 5
Optimal architecture from BPNN trial and error optimization. This figure shows the result of the BPNN trial and error procedure on one data set from each epistasis model. This shows the NN architecture for the best classification error selected from Table 1.
Similar articles
- Comparison of approaches for machine-learning optimization of neural networks for detecting gene-gene interactions in genetic epidemiology.
Motsinger-Reif AA, Dudek SM, Hahn LW, Ritchie MD. Motsinger-Reif AA, et al. Genet Epidemiol. 2008 May;32(4):325-40. doi: 10.1002/gepi.20307. Genet Epidemiol. 2008. PMID: 18265411 - A review for detecting gene-gene interactions using machine learning methods in genetic epidemiology.
Koo CL, Liew MJ, Mohamad MS, Salleh AH. Koo CL, et al. Biomed Res Int. 2013;2013:432375. doi: 10.1155/2013/432375. Epub 2013 Oct 21. Biomed Res Int. 2013. PMID: 24228248 Free PMC article. - GPNN: power studies and applications of a neural network method for detecting gene-gene interactions in studies of human disease.
Motsinger AA, Lee SL, Mellick G, Ritchie MD. Motsinger AA, et al. BMC Bioinformatics. 2006 Jan 25;7:39. doi: 10.1186/1471-2105-7-39. BMC Bioinformatics. 2006. PMID: 16436204 Free PMC article. - Artificial intelligence to deep learning: machine intelligence approach for drug discovery.
Gupta R, Srivastava D, Sahu M, Tiwari S, Ambasta RK, Kumar P. Gupta R, et al. Mol Divers. 2021 Aug;25(3):1315-1360. doi: 10.1007/s11030-021-10217-3. Epub 2021 Apr 12. Mol Divers. 2021. PMID: 33844136 Free PMC article. Review. - Siamese Neural Networks: An Overview.
Chicco D. Chicco D. Methods Mol Biol. 2021;2190:73-94. doi: 10.1007/978-1-0716-0826-5_3. Methods Mol Biol. 2021. PMID: 32804361 Review.
Cited by
- Interaction models matter: an efficient, flexible computational framework for model-specific investigation of epistasis.
Batista S, Madar VS, Freda PJ, Bhandary P, Ghosh A, Matsumoto N, Chitre AS, Palmer AA, Moore JH. Batista S, et al. BioData Min. 2024 Feb 28;17(1):7. doi: 10.1186/s13040-024-00358-0. BioData Min. 2024. PMID: 38419006 Free PMC article. - Gamma-Aminobutyric Acid Type A Receptor Variants are Associated with Autism Spectrum Disorders.
Adak P, Banerjee N, Sinha S, Bandyopadhyay AK. Adak P, et al. J Mol Neurosci. 2023 May;73(4-5):237-249. doi: 10.1007/s12031-023-02113-2. Epub 2023 Mar 21. J Mol Neurosci. 2023. PMID: 36943547 - Identification of Clinically Relevant HIV Vif Protein Motif Mutations through Machine Learning and Undersampling.
Altamirano-Flores JS, Alvarado-Hernández LÁ, Cuevas-Tello JC, Tino P, Guerra-Palomares SE, Garcia-Sepulveda CA. Altamirano-Flores JS, et al. Cells. 2023 Feb 28;12(5):772. doi: 10.3390/cells12050772. Cells. 2023. PMID: 36899908 Free PMC article. - Multifactor dimensionality reduction reveals the effect of interaction between ERAP1 and IFIH1 polymorphisms in psoriasis susceptibility genes.
Zhang C, Qin Q, Li Y, Zheng X, Chen W, Zhen Q, Li B, Wang W, Sun L. Zhang C, et al. Front Genet. 2022 Nov 8;13:1009589. doi: 10.3389/fgene.2022.1009589. eCollection 2022. Front Genet. 2022. PMID: 36425068 Free PMC article. - learnMET: an R package to apply machine learning methods for genomic prediction using multi-environment trial data.
Westhues CC, Simianer H, Beissinger TM. Westhues CC, et al. G3 (Bethesda). 2022 Nov 4;12(11):jkac226. doi: 10.1093/g3journal/jkac226. G3 (Bethesda). 2022. PMID: 36124944 Free PMC article.
References
- Templeton AR. Epistasis and complex traits. In: Wade M, Brodie III B, Wolf J, editor. Epistasis and Evolutionary Process. Oxford, Oxford University Press; 2000.
- Moore JH, Williams SM. New strategies for identifying gene-gene interactions in hypertension. Ann Med. 2002;34:88–95. - PubMed
- Bellman R. Adaptive Control Processes. Princeton, Princeton University Press. 1961.
- Bhat A, Lucek PR, Ott J. Analysis of complex traits using neural networks. Genet Epidemiol. 1999;17:S503–S507. - PubMed
- Curtis D, North BV, Sham PC. Use of an artificial neural network to detect association between a disease and multiple marker genotypes. Ann Hum Genet. 2001;65:95–107. - PubMed
Publication types
MeSH terms
Grants and funding
- HL65962/HL/NHLBI NIH HHS/United States
- U19 HL065962/HL/NHLBI NIH HHS/United States
- U01 HL065962/HL/NHLBI NIH HHS/United States
- R01 AG020135/AG/NIA NIH HHS/United States
- GM31304/GM/NIGMS NIH HHS/United States
- AG20135/AG/NIA NIH HHS/United States
- AG19065/AG/NIA NIH HHS/United States
- P01 GM031304/GM/NIGMS NIH HHS/United States
- R01 HL065234/HL/NHLBI NIH HHS/United States
- HL65234/HL/NHLBI NIH HHS/United States
- LM007450/LM/NLM NIH HHS/United States
- R01 AG019085/AG/NIA NIH HHS/United States
- T15 LM007450/LM/NLM NIH HHS/United States
LinkOut - more resources
Full Text Sources