DNA arrays have been applied to gene expression studies in bacteria, plants, yeast and mammals. Disease-related changes in gene expression have been identified in cancer, inflammatory diseases and heart diseases. Array analyses have also been used for identification of new drug targets and mechanisms of drug action.
Two types of DNA arrays have been utilized for the profiling of gene expression: cDNA arrays and oligonucleotide arrays. In addition, arrays differ with respect to methods of arraying, chemistry, linkers, hybridization and detection. In a cDNA array, cDNA fragments usually produced by PCR are spotted to microscope slides (microarrays) or nylon membranes (macroarrays) [23, 24]. Oligonucleotide arrays developed by Affymetrix are composed of thousands of oligonucleotides, 25 nucleotides in length . mRNAs from samples of interest are labelled with fluorescent dyes (Cy3 and Cy5) or radioactive nucleotides (33-P) and hybridized with immobilized targets. Two-colour fluorescent detection can be used with glass mi-croarrays. Arrays with radioactive detection are more sensitive than fluorescent arrays, requiring only 0.5 ^g of mRNA in contrast to 2-5 ^g of mRNA needed for arrays with fluorescent detection. After hybridisation, signal intensities are measured with confocal fluorescent microscopy or phosphoimager, and special software is used for the rapid identification of differentially expressed genes. Clustering methods can be used to find patterns in differentially expressed genes .
Comparison of expression data from multiple arrays requires normalization. Use of signal intensities of a subset of genes e.g. housekeeping genes or commercial hybridization controls can be used for the normalization for divergent samples. If intensities of all genes are used, samples have to be closely related . To describe the difference in signal intensities and thus in expression, intensity ratios have to be calculated. We have developed a formula where signals are normalized and intensity scores (fold increase or decrease) are calculated
144 I 9 Genes Involved in Atherosclerosis int GDA1 + n m x int GDA2 + n
Where the int GDA 1 and GDA 2 are intensities of filters 1 and 2, m is the average of all intensities on filter 1 divided by the average of intensities on filter 2 and n is 0.2 x m. The rationale for using the formula was to avoid false results caused by very low signal intensities which with the current formulation only produce values ~ 1. The formula also takes into account possible differences in the general background of the filters. Genes showing > 1.5-2 - fold increase or decrease should be processed further. Statistical significances of the differences are calculated according to Claverie .
Gene expression studies produce similar needs to use bioinformatics irrespective of the array method used. The enormous amount of data has to be analyzed and presented in a meaningful way, which has been identified as the greatest challenge in the array research. Computational biology and mathematical modelling need to be integrated with DNA array related work. Ideally, data from gene expression experiments require a uniform format so that the results can also be used for meta-analyses. The enormous amount of data may be most sophistically published through www interface. One potential alternative would be the establishment of a centralized public data bank having a similar organization as e.g.
Was this article helpful?