Proceedings of the XLV Italian Society of Agricultural Genetics - SIGA Annual Congress

Salsomaggiore Terme, Italy - 26/29 September, 2001

ISBN 88-900622-1-5

Poster Abstract

ASSESSING SOURCES OF VARIABILITY IN ARABIDOPSIS MICROARRAYS

ACCERBI M., LANDGRAF J., SCHAFFER R., WISMAN P.

MSU-DOE Plant Research Laboratory, Michigan State University, East Lansing, MI, 48823-1312, U.S.A

Many researchers are turning to microarrays to measure global changes in gene expression. The Arabidopsis Functional Genomic Consortium has constructed a microarray containing about 11,000 non-redundant ESTs from the Newman collection representing approximately 8,000 genes. As microarrays are a relatively new field, standardized protocols are still being developed. Every procedure in the microarray methodology is a potential source of fluctuation, leading to noise in the whole system. In addition, finding meaningful differences between the two samples often uses arbitrary cutoff values.

To address these issues, a series of negative control experiments using the same total RNA labeled with Cy3 or Cy5 was hybridized to 6 microarrays. We found the amount of noise generated is related to the quality of the slides, and multiple repetitions are necessary to compensate for the technical variation. We have also investigated the differences between labeling from total RNA and from polyA RNA derived from the same batch of total RNA. Comparisons of total/polyA yielded similar results, however some small differences between the methods were observed. Using stringent spot quality selection criteria, approximately 6,500 ESTs from 6 repetitions were analyzed. Of these, about 1.5% showed a ratio greater than 2, which is a cutoff commonly used. This indicates that for many experiments total RNA may be sufficient to identify most expression changes. Overall our results suggest that careful consideration is needed in planning, conducting and analyzing microarray experiments.