Improved procedures to mine data obtained from spotted cDNA arrays.

We studied six types of errors that affect the experimental results obtained with the spotted cDNA technology, proposing a new experimental strategy and a four-step algorithm to improve data accuracy. In an experiment analyzed with this method, mRNA extracted from neuroblastoma (N2A) cells and a clone transfected with the neuronal gap junction protein Cx36 were hybridized on 10 chips, each spotted with 2512 known mouse cDNAs. We found 53 upregulated and 23 downregulated genes with expression ratios exceeding 1.5-fold. By contrast, the newly introduced standard gene expression score, which considers the normal variability of gene expression in the reference cells, revealed that Cx36 transfection (p < 0.05) upregulated 161 genes and downregulated 61 genes. These results demonstrate the necessity to correct the raw data obtained from cDNA microarrays for systematic errors and indicate that use of the standard expression score significantly extends the identification of gene expression changes following experimental intervention.