English
 
User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Proceedings

Gaussian mixture density estimation applied to microarray data

MPS-Authors
/persons/resource/persons50571

Steinhoff,  Christine
Dept. of Computational Molecular Biology (Head: Martin Vingron), Max Planck Institute for Molecular Genetics, Max Planck Society;

Müller,  Tobias
Max Planck Society;

/persons/resource/persons50450

Nuber,  Ulrike A.
Dept. of Human Molecular Genetics (Head: Hans-Hilger Ropers), Max Planck Institute for Molecular Genetics, Max Planck Society;

/persons/resource/persons50613

Vingron,  Martin
Gene regulation (Martin Vingron), Dept. of Computational Molecular Biology (Head: Martin Vingron), Max Planck Institute for Molecular Genetics, Max Planck Society;

Locator
There are no locators available
Fulltext (public)
There are no public fulltexts available
Supplementary Material (public)
There is no public supplementary material available
Citation

Steinhoff, C., Müller, T., Nuber, U. A., & Vingron, M. (2003). Gaussian mixture density estimation applied to microarray data. Berlin [et al]: Springer.


Cite as: http://hdl.handle.net/11858/00-001M-0000-0010-8B3C-2
Abstract
Several publications have focused on fitting a specific distribution to overall microarray data. Due to a number of biological features the distribution of overall spot intensities can take various shapes. It appears to be impossible to find a specific distribution fitting all experiments even if they are carried out perfectly. Therefore, a probabilistic representation that models a mixture of various effects would be suitable. We use a Gaussian mixture model to represent signal intensity profiles. The advantage of this approach is the derivation of a probabilistic criterion for expressed and non-expressed genes. Furthermore our approach does not involve any prior decision on the number of model parameters. We properly fit microarray data of various shapes by a mixture of Gaussians using the EM algorithm and determine the complexity of the mixture model by the Bayesian Information Criterion (BIC). Finally, we apply our method to simulated data and to biological data.