Skip to main content
ARS Home » Research » Publications at this Location » Publication #132187
Title: ANALYSIS OF DNA MICROARRAY DATA

Author
item ALKHAROUF, NADIM - GEORGE MASON UNIV.
item CHOUIKHA, IMED - GEORGE MASON UNIV.
item Matthews, Benjamin - Ben

Submitted to: BARC Poster Day
Publication Type: Abstract Only
Publication Acceptance Date: 4/18/2002
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Analysis of DNA microarrays involves the extraction of fluorescent intensity from raw image files generated by the scanner, storing the extracted data in a database, normalizing the data, conducting statistical analysis and finally querying the analyzed data to find biologically meaningful results. Here we describe the analysis of a microarray experiment conducted at the Beltsville Agricultural Research Center (BARC) where the goal was to identify resistance genes against the soybean cyst nematode (SCN). We describe the different image processing tools that are used and compare the freely available ScanAlyze with the commercial and more sophisticated SPOT software. SPOT provides more accurate results due to it's advanced algorithm that can automatically find printed spots on a slide and find their exact dimensions. Our data is stored and analyzed in SGMD, the Soybean Genomics and Microarray Database at BARC. Different normalization methods, including whole slide normalization, control normalization, regression, and Lowess, were compared by applying one way analysis of variance (ANOVA) tests on replicated slides that were printed and hybridized with the same probes at the same time. ANOVA was applied on a gene-by-gene basis, since four replicates for each gene were printed on each slide. We found that whole slide and control normalization did not reduce the variability across slides since only 50-60% of the genes passed the ANOVA. Regression normalization faired better, where almost 70-75% of the genes passed ANOVA. Applying Lowess on a grid-by-grid basis provided the best overall result, where greater than 90% of the genes passed ANOVA and thereby showed similar expression across slides. Thus Lowess reduces the intra- and inter-array variability observed in microarray experiments, which is of paramount importance if we are to compare and correlate different experiments. Our web site provides user interfaces for scientists to query and analyze their results and our web site provides results obtained from our soybean-SCN studies and results from microarray experiments conducted by other BARC scientists that were analyzed by us and stored in SGMD.