Introduction
Research in the past decade has revealed that microRNAs (miRNAs) are widespread and that they are likely to underlie an appreciably larger set of disease processes than is currently known. The first miRNAs and their functions were determined via classical genetic techniques. Soon after, a number of miRNAs were discovered experimentally (Lagos-Quintana et al., 2001). However, the characterization of miRNA function remained elusive owing to low-throughput experiments and often indeterminate results, most notably for those miRNAs which have multiple roles in multiple tissues. High-throughput experimental methods for miRNA target identification are the ideal solution, but such methods are not currently available. As a result, computational methods were developed, and are still regularly used, for the purpose of identifying miRNA targets.
Most current target prediction programs require the sequences of known miRNAs. Currently, there are 332 known miRNAs in the human genome. The estimation of the total number of miRNAs varies from publication to publication (Lim et al., 2003; Bentwich et al., 2005). In a recent paper, Bentwich et al. contended that there are at least 500 more miRNAs that are yet to be identified (Bentwich et al., 2005). Despite the number of unknown miRNAs, computational approaches based on features of known miRNAs have been instrumental in the discovery of as-of-yet-unknown miRNAs in the genome. The past few years have witnessed an explosion in information regarding the genomic organization of miRNAs, the biogenesis of miRNAs, the targeting mechanisms of miRNAs, and the regulatory networks in which miRNAs are involved.