Abstract
Most human cancers are characterized by genomic instability. Changes associated with such may result in altered expression of numerous genes. The sequence information available in the public databases can be used to identify transcripts differentially expressed in cancers. Determining cancer-related genes that are commonly deregulated in different tumor types may facilitate identification of targets for cancer diagnoses and therapeutic treatments. Using a data-mining tool named Digital Differential Display (DDD) from the UniGene database at the NCBI web site, gene expression levels of ten different tumor types and their counterpart normal tissues were analyzed. Unigenes which showed transcriptional regulation in more than five tumor types with ≧2-fold differences from normal tissues were identified. The expression data of selected Unigenes were subjected to clustering analysis. 127 commonly up-regulated genes and 92 commonly down-regulated genes were identified. Clustering analysis using these genes showed that most tumor types can be clustered into a separate branch from most normal tissues. Nineteen genes that have been shown to be involved in carcinogenesis by experimental evidence were also identified. Present computational analyses revealed 219 candidate cancer-related genes that are commonly deregulated in ten human tumor types which may contribute to the progress of carcinogenesis.