Abstract:

Grid Computing technologies enable large-scale aggregation and sharing of untapped computational, data, and other resources distributed across institutional boundaries. Harnessing these new technologies effectively will transform key scientific disciplines such as life sciences and pave the way towards tackling complex biomedical problems. One such challenge is the identification of viral-encoded microRNAs and human mRNA targets interactions. MicroRNAs (miRNAs) constitute an ubiquitous class of small (about 21–23 nucleotides) non-protein-coding RNAs (ncRNAs) regulating gene expression in diverse biological processes including developmental timing, signal transduction, apoptosis, cell proliferation, and tumorigenesis across plants, animals, and even viruses. Since high-throughput experimental methods for identifying viral-encoded miRNAs and human mRNA targets are not yet available, few experimentally validated viral-encoded miRNAs and even fewer for viral/human miRNAs-target interactions have been reported. This bleak situation results in many unanswered questions concerning the detailed mechanism of how viral-encoded miRNAs are involved in disabling/hijacking the host defense including the degree of down-regulating gene expression. As part of our research effort, we have developed a set of Perl-based bioinformatics tools (known as miRnalzer) for the identification of viral-encoded miRNAs. It was estimated that three years on a single machine is required to analyze the entire 6,656 native RNA sequences and their corresponding 266,240,000 synthetic RNA sequences. With the hope to considerably shorten the computation time (to about three weeks) and to maximize available untapped computational resources usage, the miRnalzer was migrated to a Grid Computing environment known as the BioGrid. The experimental results and performances of miRnalzer conducted on BioGrid are presented and discussed.

Keywords: microRNAs; viruses; de novo identification; Grid Computing;