Rna22 is a pattern-based algorithm for the discovery of microRNA target sites and the corresponding heteroduplexes.

The algorithm is conceptually distinct from other methods for predicting microRNA:mRNA heteroduplexes in that it does not use experimentally validated heteroduplexes for training, instead relying only on the sequences of known mature miRNAs that are found in the public databases. The key idea of rna22 is that the reverse complement of any salient sequence features that one can identify in mature microRNA sequences (using pattern discovery techniques) should allow one to identify candidate microRNA target sites in a sequence of interest: rna22 makes use of the Teiresias algorithm to discover such salient features. Once a candidate microRNA target site has been located, the targeting microRNA can be identified with the help of any of several algorithms able to compute RNA:RNA heteroduplexes. A new version (v2.0) of the algorithm is now available: v2.0-beta adds probability estimates to each prediction, gives users the ability to choose the sensitivity/specificity settings on-the-fly, is significantly faster than the original, and can be accessed through http://cm.jefferson.edu/rna22/Interactive/.