Statistical adjustment of signal censoring in gene expression experiments

Wit, E. and McClure, J. (2003) Statistical adjustment of signal censoring in gene expression experiments. Bioinformatics, 19, pp. 1055-1060. (doi: 10.1093/bioinformatics/btg003)

Abstract

Motivation: Numerical output of spotted microarrays displays censoring of pixel intensities at some software dependent threshold. This reduces the quality of gene expression data, because it seriously violates the linearity of expression with respect to signal intensity. Statistical methods based on typically available spot summaries together with some parametric assumptions can suggest ways to correct for this defect Results: A maximum likelihood approach is suggested together with a sensible approximation to the joint density of the mean, median and variance—which are typically available to the biological end-user. The method ‘corrects’ the gene expression values for pixel censoring. A by-product of our approach is a comparison between several two-parameter models for pixel intensity values. It suggests that pixels separated by one or two other pixels can be considered independent draws from a Lognormal or a Gamma distribution.

• Item Type: Articles
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: UNSPECIFIED
• Authors: Wit, E., and McClure, J.
• College/School: College of Science and Engineering > School of Mathematics and Statistics > Statistics
• Journal Name: Bioinformatics