水稻MicroRNA的预测及实验验证

根据已报道水稻pre-miRNA的序列与结构信息，利用支持向量机（support vector machine, SVM）方法在miRNA前体上预测成熟区，产生一个模型——mature-SVM.它预测水稻成熟区的敏感性和特异性分别为86.7% 和100%；然后,用这个模型对从水稻基因组中筛选出的46.501条pre-miRNA进行成熟链预测，此外再根据miRNA的作用原理用blast程序所进一步的筛选，得到了127条pre-miRNA及成熟miRNA；除去其中已知的21条，最后得到106条候选的新的水稻miRNA. 从中随机挑取10条进行Northern验证，结果有4条miRNA得到确认.

Prediction and Validation of MicroRNAs from Rice Genome Using Mature-SVM

MicroRNAs(miRNAs), ranging in size from 20～25 nt, are a growing family of noncoding RNAs that play important roles in regulation of expression of target genes. Hundreds of miRNAs have been identified by experimental complementary DNA cloning and computational methods. However, these available approaches could only detect abundantly expressed miRNAs or close homologs of known miRNAs. Ab initio method for distinguishing mature region on pre-miRNAs is lacking. Identifying mature regions on pre-miRNAs are important both for further understanding of miRNAs and for developing ab initio prediction methods that can discover new miRNAs without known homology. A set of novel features of local structure and sequence information were proposed for distinguishing mature region on pre-miRNAs using support vector machine(SVM). Mature-SVM, a new method applied to predict mature region on the miRNA precursors, sensitivity of 86.7% and specificity of 100% on rice data set,were obtained. Then the mature-SVM was applied for predicting mature region on 46 501 pre-miRNAs which were predicted from rice genome according to the characteristics of the known miRNAs. Finally, 106 novel miRNA candidates were obtained from rice genome. Ten miRNA candidates selected randomly from the 106 candidates were confirmed by Northern blotting and 4 candidates were validated with an accuracy 40%. The successful ab initio detection of mature region on the precursors opens a new way for discovering miRNAs.