Use of tiling array data and RNA secondary structure predictions to identify noncoding RNA genes

We have combined tiling array data with genome wide structural RNA predictions to search for novel noncoding and structural RNA genes that are expressed in the human neuroblastoma cell line SK-N-AS. Using this strategy, we identify thousands of human candidate RNA genes. To further verify the expression of these genes, we focused on candidate genes that had a stable hairpin structures or a high level of covariance. Using northern blotting, we verify the expression of 2 out of 3 of the hairpin structures and 3 out of 9 high covariance structures in SK-N-AS cells.Our results demonstrate that many human noncoding, structured and conserved RNA genes remain to be discovered and that tissue specific tiling array data can be used in combination with computational predictions of sequences encoding structural RNAs to improve the search for such genes.The sequencing of the human genome marked the starting point of a very difficult task: to make sense of the enormous amount of information stored in the genome by annotating the functionally important regions. Emphasis was initially put on the protein coding DNA sequences, which are generally well conserved and can easily be converted into the corresponding protein sequence. However, in recent years it has become clear that large parts of the noncoding DNA present in the human genome is functional and that noncoding genes may be as abundant as protein coding genes [1].Central to this realization has been the sequencing of additional mammalian genomes. Comparative genomics have demonstrated that the fraction of the human genome that is under purifying selection is much larger than the part that makes up the protein coding sequence, suggesting that many non protein coding regions of the genome have important functions [2]. Conserved sequence elements in promoter, intron and untranslated regions (UTRs) control transcription and processing of mRNAs [3]. Moreover, distant enhancer elements also influence transcription over long dista