You are here : Home > Scientific news > DSIR: A new in silico method for predicting siRNA efficacy

Highlight | Bioinformatic

DSIR: A new in silico method for predicting siRNA efficacy


​​​​

Researchers at our institute have teamed up with bioinformatics researchers at the Ecole des Mînes in Paris to devise a new method for predicting siRNA efficacy. DSIR (Designer of Small Interfering Rna).​​​

Published on 1 February 2007

Exogenous small interfering RNA (siRNA) provides a powerful molecular tool for targeted silencing of gene expression and is widely used in functional gene studies or in the identification of new drug targets. While there have been major recent advances in our understanding of the molecular mechanisms underlying the small interfering RNA pathway, we still lack a satisfactory set of design rules for producing efficient siRNA.

As part of a programme backed by the National League Against Cancer, researchers at our institute have teamed up with bioinformatics researchers at the Ecole des Mînes in Paris to devise a new method called DSIR (Designer of Small Interfering Rna) for predicting siRNA efficacy based on a simple linear model. This statistical model, trained and tested on a large set of siRNA sequences recently published by Huesken et al., 2005, Nat. Methods, provides more accurate predictions of potential siRNA efficacy with the added advantage of being directly interpretable in terms of biological descriptors. Analysis of this linear model can therefore be used to detect and quantify the effect of both previously and newly identified nucleotide preferences; this model also reveals the presence of asymmetrical units in functional siRNA sequences that contain additional information on nucleotide preferences at a given site. Very recently, a comparative study ranked DSIR as the best performer among currently existing algorithms [Matveeva et al., 2007, Nucleic Acid Res.]. DSIR is now available as a Web tool providing a secure access to the predictive method together with an off-target detection algorithm.

Top page