|Schnell ii, Raymond|
Submitted to: Biotechniques
Publication Type: Peer reviewed journal
Publication Acceptance Date: 11/1/2003
Publication Date: 2/1/2004
Citation: Brown, J.S., Kuhn, D., Wisser, R.J., Power, E.J., Schnell II, R.J. 2004. Determination and relative quantification of sources of variation in a replicated microarray experiment using resistance gene homologue sequences. Biotechniques. Interpretive Summary: Microarray technology has become of the most important tools in the study of gene expression. In a cDNA microarray study, sequences of DNA, usually part of expressed genes, are spotted onto an array slide in a known pattern. About 3,000 of these cDNA sequences can be placed on a single slide. To study gene expression, a plant is placed in a treatment condition, such as exposure to a disease. Plant m-RNA is then isolated from both treated and non-treated plants, and labeled with a fluorescent dye. The m-RNA is then hybridized with the DNA on the slide and the dye is excited by a laser. The intensity of the fluorescence from the dye gives a measure of the copy number of the m-RNA, and hence a measure of gene expression. The object of this study was to validate and test the accuracy our microarrayer's output. The amount of replication required to have confidence in the data was especially of interest. Known Resistant Gene Homologues (RGH), a type of plant disease resistance gene, were used which had been isolated and sequenced in our lab from Theobroma cacao. The RGH were spotted onto the array slide in a precise statistical design. We then used two of the same RGH as probes, simulating mRNA, to test the system. The two probes were labeled and hybridized to the slides. We were please to find that the microarray system was working very well with little variation due to the mechanics of placing the DNA on the slide, to slide to slide variation, or position of the cDNA spot on the slide. This is one of first reports using a sophisticated statistical design to quantify variation in microarray experiments.
Technical Abstract: Microarrays of cDNA provide effective methodology for analysis of gene expression, giving data that becomes integrated into large databases. The integrity, accuracy, and precision of data at each step becomes critical, and error can come from many sources. Errors at one point in the research process can have devastating ramifications on resulting conclusions. It seems therefore prudent when beginning microarray research to validate the proces. Prior knowledge of relative sources of variation is also useful for effective experimental design and attaining desired statistical power. Our goal was to determine relative sizes of contributions to variance by carrying out an exploratory validation experiment using a split plot experimental design, which used 67 related sequence fragments from candidate Resistance Gene Homologues (RGH) isolated from Theobroma cacao L Two of the RGH's were also used in the two fluorescent dye channels (Cy3 and Cy5) in the probe solution. Variability was then quantified by sources: slides, quadrants, pins, replications, RGH (nested by group), and between dyes and RGH fragments in the dye, as well as interactions. Largest sources of variability were due to probe sequences and RGH (24.97% and 21.64%, respectively) followed by Dye and RGH group effects (11.65% and 4.70%). Slide, quadrant, and pin effects were not significant, though replications were highly significant, as reported elsewhere. Overall %CV based on experimental error was 4.80, indicating acceptable precision. We were able to separate expressed signal from background consistently, and able to discern the expression of RGH related to those in the probe solution from those less related, giving us confidence in the system.