Informatic selection of a neural crest-melanocyte cDNA set for microarray analysis

With cDNA microarrays, it is now possible to compare the expression of many genes simultaneously. To maximize the likelihood of finding genes whose ex pression is altered under the experimental conditions, it would be advantag eous to be able to select clones for tissue-appropriate cDNA sets. We have taken advantage of the extensive sequence information in the dbEST expresse d sequence tag (EST) database to identify a neural crest-derived melanocyte cDNA set for microarray analysis. Analysis of characterized genes with dbE ST identified one library that contained ESTs representing 21 neural crest- expressed genes (library 198). The distribution of the ESTs corresponding t o these genes was biased toward being derived from library 198. This is in contrast to the EST distribution profile for a set of control genes, charac terized to be more ubiquitously expressed in multiple tissues (P < 1 x 10(- 9)). From library 198, a subset of 852 clustered ESTs were selected that ha ve a library distribution profile similar to that of the 21 neural crest-ex pressed genes. Microarray analysis demonstrated the majority of the neural crest-selected 852 ESTs (Mel1 array) were differentially expressed in melan oma cell lines compared with a non-neural crest kidney epithelial cell line (P < 1 x 10(-8)). This was net observed with an array of 1,238 ESTs that w as selected without library origin bias (P = 0.204). This study presents an approach for selecting tissue-appropriate cDNAs that can be used to examin e the expression profiles of developmental processes and diseases.