MicroRNAs, which are small ribonucleic acids that bind to 3'UTR regions of mRNAs by base complementation, play crucial roles in regulation of development and differention . Herein we report on development of a web interfaced database, Bilkent University miRNA Sequence and Expression database [Figure 1; http://126.96.36.199/~koray webcite] that integrates the species-specific mature miRNA sequence information and a set of associated public microarray data as tabular and graphical summaries suplemented with statistical analyses http://www.bioconductor.org webcite. The database also makes use of GO annotation data of miRNAs targets to determine the significance of GO term enrichment in a subset of miRNAs.
Figure 1. The interface of miRNA sequence and expression database.
Data and methods
Bilkent University miRNA sequence and expression database was constructed using Mysql version 14.7 on Suse Linux 10.0 server; the web interface implemented in HTML 4.0 combined with PHP version 4.4.0. Statistical calculations were performed using R package version 2.2.1. miRNA mature sequences from Homo sapiens, Mus musculus, Danio rerio and Caenorhabditis elegans were downloaded from miRBase of Sanger Institute database http://microrna.sanger.ac.uk/ webcite version 8.2. Two independent microarray data sets [3,4] were associated with human and mouse miRNA dinucleotide motif frequency, respectively. Human miRNA targets were extracted from Argonaute Database of Heildelberg http://www.ma.uni-heidelberg.de/apps/zmf/argonaute/interface/ webcite. GO ontology data were linked with Argaonaute gene symbols and alias data from Entrez Gene Database http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?DB=gene webcite.
Results and conclusions
The database allows for selection of miRNAs based on their dinucleotide properties and reports expression pattern of this particular set of sequences (Figure 2).
Figure 2. Left: Bar graph of microarray expression data for Hsa-mir-15a and Hsa-mir-16. Right: miRNA target GO annotation distribution and significantly enriched GO terms reported. One of the annotated targets were found to be BCL-2, a well known anti-apoptotic protein .
High expression of hsa-mir-15a and hsa-mir-16, known to be deleted in chronic lymphocytic leukemia, was detected in bone marrow and spleen (Figure 2a). Analysis of these miRNAs in terms of their targets resulted in significant representation of antiapoptosis and humoral immune response GO terms, both attributable to BCL-2 (Figure 2b). Although BCL2 is well known for its role in cell survival, there is no known direct relation of BCL-2 with immune response. Future studies involve integration of multiple species-specific gene expression data sets and implementation of correspondence analysis tools for multivariate analysis of sequence and expression data. The presented database will help understand miRNAs in a systems biology context via integration of the sequence, expression, and functional attributes.