Improving model predictions for RNA interference activities that use support vector machine regression by combining and filterin

Background: RNA interference (RNAi) is a naturally occurring phenomenon that results in the suppression of a target RNA sequence utilizing a variety of possible methods and pathways. To dissect the factors that result in effective siRNA sequences a regression kernel Support Vector Machine (SVM) approach was used to quantitatively model RNA interference activities. Results: Eight overall feature mapping methods were compared in their abilities to build SVM regression models that predict published siRNA activities. The primary factors in predictive SVM models are position specific nucleotide compositions. The secondary factors are position independent sequence motifs (N-grams) and guide strand to passenger strand sequence thermodynamics. Finally, the factors that are least contributory but are still predictive of efficacy are measures of intramolecular guide strand secondary structure and target strand secondary structure. Of these, the site of the 5' most base of the guide strand ...

• Type: Journal
• Year: 2007
• Where: BMCBI
• Authors: Andrew S. Peek
• Keywords: BMCBI 2007, Feature Mapping Methods, Predictive, SVM Regression Models