Background adjustment of cDNA microarray images by Maximum Entropy distributions

Authors:
Christos Argyropoulos;Antonis Daskalakis;George C. Nikiforidis;George C. Sakellaropoulos
Affiliations:
Department of Medical Physics, School of Medicine, University of Patras, GR-26504 Rion, Greece and Renal and Electrolyte Division, University of Pittsburgh Medical Center, A919 3550 Terrace Street ...;Department of Medical Physics, School of Medicine, University of Patras, GR-26504 Rion, Greece;Department of Medical Physics, School of Medicine, University of Patras, GR-26504 Rion, Greece;Department of Medical Physics, School of Medicine, University of Patras, GR-26504 Rion, Greece
Venue:
Journal of Biomedical Informatics
Year:
2010

Citing 6
Cited 1

Hausdorff moment problem and maximum entropy: a unified approach

Applied Mathematics and Computation
Analysis of variance components in gene expression data

Bioinformatics
MAGIC Tool: integrated microarray data analysis

Bioinformatics
A comparison of algorithms for maximum entropy parameter estimation

COLING-02 proceedings of the 6th conference on Natural language learning - Volume 20
Donuts, scratches and blanks: robust model-based segmentation of microarray images

Bioinformatics
Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropy

IEEE Transactions on Information Theory

Evaluating the effect of various background correction methods regarding noise reduction, in two-channel microarray data

Computers in Biology and Medicine

Quantified Score

Hi-index	0.01

Visualization

Abstract

Many empirical studies have demonstrated the exquisite sensitivity of both traditional and novel statistical and machine intelligence algorithms to the method of background adjustment used to analyze microarray datasets. In this paper we develop a statistical framework that approaches background adjustment as a classic stochastic inverse problem, whose noise characteristics are given in terms of Maximum Entropy distributions. We derive analytic closed form approximations to the combined problem of estimating the magnitude of the background in microarray images and adjusting for its presence. The proposed method reduces standardized measures of log expression variability across replicates in situations of known differential and non-differential gene expression without increasing the bias. Additionally, it results in computationally efficient procedures for estimation and learning based on sufficient statistics and can filter out spot measures with intensities that are numerically close to the background level resulting in a noise reduction of about 7%.