With a limited number of biological replicates in most microarray studies due to cost considerations, it is not clear how well these transcriptomics-based gene classifiers will perform in validations.
However, power and sample size have been viewed as complicated and difficult issues for microarray studies due to the large number of genes being investigated and little knowledge of the degree of natural expression variation within a population.
Of the total of 14,811 clones on the chips, 9376 clones that were filtered out for further analyses in the regular cDNA microarray studies due to the low hybridization intensity (intensities <500 unit after the noise subtracted) were selected for examination of their hybridization intensity in the SSH/microarray assays.
The benefit of excluding poor quality arrays in differential expression analyses was further demonstrated through statistical power calculations for these two clinical trials.Statistical power is a major limiting factor in clinical microarray studies due to limited sample size and lack of technical replicates.
These genes and their downstream effectors were not identified in the current microarray study, likely due to many differences in experimental design between the two studies (e.g., pathogen exposure levels, tissue types, water temperatures, age of fish at exposure, etc).
However, strong down-regulation of major intrinsic proteins in short fiber mutants was not noticed before in our microarray studies, probably due to limitations of microarray techniques.
Head kidney tissues, rather than those of the injection site (pylorus/pancreas) were used for the screening of gene transcripts by microarrays in this study due to the difficulty in obtaining homogeneity in the samples of the latter as the organs (pyloric caeca and pancreas) are anatomically intertwined.
It is unclear why there was a delay in detection upregulated genes in our qRT-PCR study relative to our microarray study, but may be due to differences in the sex and stains of mice used in the studies.
However, no consistent conclusion could be drawn from most of the previous microarray-based studies due to the limitations of inter-platform differences and the relatively small sample sizes investigated [10], [12], [17], [18].
ANNs were chosen as the bioinformatics tool for microarray data analysis for the present study due to their ability to cope with complex data and the potential for modelling data of high nonlinearity.
