multiscan: Combining multiple laser scans of microarrays

Bruce Worton, M. R. Khondoker, C. A. Glasbey

Research output: Non-textual formSoftware

Abstract

The sensitivity level of microarray scanners is adjustable and plays a crucial role in getting reliable measurement of the fluorescence intensity. A change in scanner setting transforms the intensity measurements by a multiplicative constant. A scanner’s sensitivity has to be raised to a certainlevel to ensure that the intensity levels of weakly expressed genes exceed the intrinsic noise level of the scanner and so become measurable. This may, however, cause another problem: signal censoring for highly expressed genes. Scanners cannot record pixel intensities above some software dependent threshold (216 − 1 = 65535, for a 16-bit computer storage system), so highly expressed genes can have pixel values which are right censored at the largest possible value that the scanner software allows. It is not usually possible to find a scanner setting which is optimal for both weakly and highly expressed genes. So, it seems reasonable to consider multiple scanning of the same microarray at different scanner settings and estimate spot intensities from these combined data. The multiscan package implements the method of Khondoker et al. (2006) for estimating gene expressions from multiple laser scans of hybridised microarrays. The method is based on a non-
linear functional regression model with both additive and multiplicative error terms. Maximum likelihood estimation based on a Cauchy distribution is used to fit the model, which reduces the sampling variability in expression estimates and is able to estimate gene expressions taking account of outliers and the systematic bias caused by signal censoring of highly expressed genes.
Original languageEnglish
Place of Publicationhttp://www.bioconductor.org
Publisher Bioconductor
Media of outputOnline
Publication statusPublished - 1 Nov 2011

Fingerprint Dive into the research topics of 'multiscan: Combining multiple laser scans of microarrays'. Together they form a unique fingerprint.

Cite this