A customized class of functions for modeling and clustering gene expression profiles in embryonic stem cells

Item Type:	Article
Title:	A customized class of functions for modeling and clustering gene expression profiles in embryonic stem cells
Creators Name:	Li, S. and Andrade-Navarro, M. and Sankoff, D.
Abstract:	Based on the trajectories of individual genes, we address the problem of clustering time course gene expression data for embryonic stem cells (ESC) differentiation. We propose a class of functions determined by only two parameters but flexible enough to model realistic time courses. This serves as a basis for a mixed model clustering method. This method takes into account (1) genetic function profile induced or controlled by other regulators, (2) unobservable random effects producing heterogeneity within gene clusters, and (3) autoregressive components defining the stochastic and autocorrelation structures. We employ an EM algorithm to fit the mixture model and clustering follows monitoring via Bayesian posterior probabilities. Our method is applied to a mouse ESC line during the first 24 hours of differentiation period. We assess the biological credibility of the results by detecting significantly associated FatiGO Gene Ontology terms for each cluster.
Keywords:	Bioactivity, Biocommunications, Bioinformatics, Biotechnology, Cells, Chlorine Compounds, Cluster Analysis, Differentiation (Calculus), Electroslag Process, Flow of Solids, Gene Expression, Information Science, Ontology, Random Processes, Risk Assessment
Source:	Lecture Notes in Computer Science
Series Name:	Lecture Notes in Computer Science
Title of Book:	Advances in bioinformatics and computational biology : third Brazilian Symposium on Bioinformatics, BSB 2008, Santo Andre, Brazil, August 28-30, 2008. Proceedings
ISSN:	0302-9743
ISBN:	978-3-540-85556-9
Publisher:	Springer
Volume:	5167
Page Range:	92-103
Date:	2008
Official Publication:	https://doi.org/10.1007/978-3-540-85557-6_9

Repository Staff Only: item control page