Email updates

Keep up to date with the latest news and content from BMC Bioinformatics and BioMed Central.

Open Access Methodology article

Nonparametric methods for the analysis of single-color pathogen microarrays

Omar J Jabado, Sean Conlan, Phenix-Lan Quan, Jeffrey Hui, Gustavo Palacios, Mady Hornig, Thomas Briese and W Ian Lipkin*

Author Affiliations

Center for Infection and Immunity Mailman School of Public Health Columbia University New York, NY USA

For all author emails, please log on.

BMC Bioinformatics 2010, 11:354  doi:10.1186/1471-2105-11-354

Published: 28 June 2010

Abstract

Background

The analysis of oligonucleotide microarray data in pathogen surveillance and discovery is a challenging task. Target template concentration, nucleic acid integrity, and host nucleic acid composition can each have a profound effect on signal distribution. Exploratory analysis of fluorescent signal distribution in clinical samples has revealed deviations from normality, suggesting that distribution-free approaches should be applied.

Results

Positive predictive value and false positive rates were examined to assess the utility of three well-established nonparametric methods for the analysis of viral array hybridization data: (1) Mann-Whitney U, (2) the Spearman correlation coefficient and (3) the chi-square test. Of the three tests, the chi-square proved most useful.

Conclusions

The acceptance of microarray use for routine clinical diagnostics will require that the technology be accompanied by simple yet reliable analytic methods. We report that our implementation of the chi-square test yielded a combination of low false positive rates and a high degree of predictive accuracy.