Overview
What is siRNA?
siRNA, also known as short-interfering RNA or silencing RNA, is a class of double-stranded RNA which participates in RNA interference (RNAi) and may be a useful tool in gene therapy.
Synthetic siRNA was discovered to be an effective means of mammalian gene suppression in 2001 by Thomas Tuschl, and since then there has been considerable interest in the area. Each strand of siRNA is relatively short--between 19-25 nucleotides long--and works by cleaving mRNA, thus disrupting the protein creation process.
The Problem
Not all strands of siRNA are equally effective, and given the sheer number of possible siRNA sequences (4n possible sequences, given a length of n nucleotides), it is important to be able to classify which siRNA sequences are functional and which are non-functional. The difficulty lies in the number of possible sequences, as well as in the fact that the mechanism which guides RNAi is not well understood. We would like to be able to construct a model or algorithm which is able to predict whether a string of siRNA is functional or not.
Previous Models
In particular, Pancoska's Eulerian graph model for predicting functional siRNA was considered (see Pancoska's Algorithm). While an interesting idea, this algorithm requires complex pattern recognition techniques in order to produce any significant results. There are also some flaws associated with the model which we take into consideration.
New Ideas
Under the premise that number of mutations to a functional strand of siRNA is directly related to the mutated strand's potential functionality, we look at the Levenshtein distance between siRNA, which counts the minimum number of substitutions and/or insertions required to go from one string to another. More about Levenshtein distance can be found on Wikipedia. Using this distance as a metric, we postulate that perhaps strings of siRNA are more likely to be functional if their Levenshtein distance from a known functional string is relatively small.
We are also looking at data concerning the "ideal" properties of functional siRNA, currently concentrating on primary structure of siRNA such as the percentage of G/C nucleotides and the ratio of A/U nucleotides at the 3' and 5' ends. We are interested in the number of strands of siRNA given these constraints. The next step is to look at the secondary structure of the siRNA, which includes the consideration of two-dimensional properties such as hairpin loops which occur in the strands.