A novel method for in silico identification of regulatory SNPs in human genome. (21st February 2017)

• Record Type: Journal Article
• Authors: Li, Rong; Zhong, Dexing; Liu, Ruiling; Lv, Hongqiang; Zhang, Xinman; Liu, Jun; Han, Jiuqiang
• Is Part Of: Journal of theoretical biology. Volume 415(2017)
• Page Start: 84
• Page End: 89
• Publication Date: 2017-02-21
• Subjects: Imbalanced data -- Hydroxyl radical cleavage patterns -- Support vector machine -- Position weight matrix; Biology -- Periodicals; Biological Science Disciplines -- Periodicals; Biology -- Periodicals; Biologie -- Périodiques; Theoretische biologie; Biology; Periodicals; 571.05
• DOI: 10.1016/j.jtbi.2016.11.022
• Languages: English
• ISSNs: 0022-5193

Abstract

Abstract: Regulatory single nucleotide polymorphisms (rSNPs), kind of functional noncoding genetic variants, can affect gene expression in a regulatory way, and they are thought to be associated with increased susceptibilities to complex diseases. Here a novel computational approach to identify potential rSNPs is presented. Different from most other rSNPs finding methods which based on hypothesis that SNPs causing large allele-specific changes in transcription factor binding affinities are more likely to play regulatory functions, we use a set of documented experimentally verified rSNPs and nonfunctional background SNPs to train classifiers, so the discriminating features are found. To characterize variants, an extensive range of characteristics, such as sequence context, DNA structure and evolutionary conservation etc. are analyzed. Support vector machine is adopted to build the classifier model together with an ensemble method to deal with unbalanced data. 10-fold cross-validation result shows that our method can achieve accuracy with sensitivity of ~78% and specificity of ~82%. Furthermore, our method performances better than some other algorithms based on aforementioned hypothesis in handling false positives. The original data and the source matlab codes involved are available athttps://sourceforge.net/projects/rsnppredict/ . Highlights: A computational method for prediction of rSNPs is proposed. A new ensemble method for handling unbalanced data is applied. Various different types of features to characterize variations are analyzed.