SUPPORT VECTOR MACHINE APPROACH FOR siRNA FUNCTIONALITY PREDICTION
Short interfering RNAs (siRNAs) are used in functional genomics to knockdown a single gene. However, only a limited fraction of siRNA appears capable of producing highly effective RNAi. So successfully prediction of siRNA efficacy is crucial for the application of the RNAi technology in practice. In this paper, we introduced a new method of siRNA efficacy prediction that used the support vector machine (SVM) algorithm together with nucleotide composition, dinucleotide composition and thermodynamic properties. The leave-one-out cross-validation evaluation of this protocol showed that we were able to predict the functionality of siRNA sequences with 75. 31％ total accuracy and 0. 50 MCC value using 43 features to describe each siRNA sequence. A backward elimination approach of possible subsets of features produced a simplified model based on 32 features that predicted at a total accuracy 80. 33％ and MCC 0. 60. Our algorithm is sensitivity and specificity.
siRNA efficacy prediction support vector machine
Peng Jiang Wei Ma Haonan Wu Jiawei Wei Fei Sang Xiao Sun Zuhong Lu
State Key Laboratory of Bioelectronics, Department of Biological Science and Medical Engineering,Southeast University, Nanjing 210096, China
国际会议
The 4th International Forum on Post-genome Technologies(4IFPT)(第四届国际后基因组生命科学技术学术论坛)
杭州
英文
391-395
2006-09-25（万方平台首次上网日期，不代表论文的发表时间）