Model the relationship between gene expression and TFBSs using a simplified neural network with bayesian variable selection

Authors:
Xiaobo Zhou;Kuang-Yu Liu;Guangqin Li;Stephen Wong
Affiliations:
HCNR-Center for Bioinformatics, Harvard Medical School, MA;HCNR-Center for Bioinformatics, Harvard Medical School, MA;Department of Neurology, The First Affiliated Hospital, Chongqing University of Medical Sciences, Chongqing, China;HCNR-Center for Bioinformatics, Harvard Medical School, MA
Venue:
ISNN'05 Proceedings of the Second international conference on Advances in Neural Networks - Volume Part III
Year:
2005

Citing 3
Cited 0

A Bayesian connectivity-based approach to constructing probabilistic gene regulatory networks

Bioinformatics
Regulatory motif finding by logic regression

Bioinformatics
Gene prediction using multinomial probit regression with Bayesian gene selection

EURASIP Journal on Applied Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Although numerous computational methods consider the identification of individual transcription factor binding sites (TFBSs), very few focus on the interactions between these sites. In this study, we study the relationship between transcription factor binding sites and microarray gene expression data. A probit regression with one linear term plus nonlinear (it is actually a simplified neural network) is used to build a predictive model of outcome of interest (either gene expression ratios or up- and down-regulations) using these transcription factor binding sites. This issue is related to the more general problem of expression prediction in which we want to find small subsets of TFBSs to be used as predictors of possible co-expressed genes and those genes do share some DNA regulatory motifs. Given some maximum number of predictors to be used, a full search of all possible predictor sets is prohibitive. This paper considers Bayesian variable selection for prediction using the nonlinear probit model (or simplified neural network). We applied this nonlinear model with Bayesian motif selection on one gene expression data set. These TFs demonstrated intricate regulatory roles either as a family or as individual members and our analysis created plausible hypotheses for combinatorial interaction among TFBSs.