| 摘要: |
| [目的]开发一个用于从pre-miRNA上识别其成熟链的miRNA预测程序miR-SVM.[方法]利用支持向量机工具,将pre-miRNA的序列结构特征作为支持向量机工具--LibSVM的输入向量,经Grid程序优化参数后,开发出一个可靠的miRNA成熟链预测程序miR-SVM.[结果]检测结果表明,在miR-SVM人数据集上得到程序的敏感性和特异性分别为83.7%和81.2%.通过杂交验证,获得ROC曲线下的面积约为87.71%,表明研究提出的序列结构特征可以有效地预测pre-miRNA成熟链.此外,用人数据训练出来的miR-SVM程序对其他20个物种的pre-miRNA成熟链进行预测,结果正确识别率为89.2%.[结论]研究开发的miR-SVM程序,成功地预测了pre-miRNA成熟链,检验结果表明,该程序具有良好的推广性,可用于miRNA试验过程中的前期预测分析. |
| 关键词: 成熟链 预测分析 支持向量机 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金，陕西省自然科学基金，浙江省生物医学工程重中之重开放基金，西北农林科技大学人才专项基金 |
|
| Idetification of mature microRNA on the precursor using ab initio prediction method |
|
|
| Abstract: |
| 【Objective】 The research predicted mature strand on the miRNA precursor.【Method】 A program,miR SVM,was developed based on support vector machine for identifying mature strand on pre miRNA.To optimize the SVM classifier,the penalty parameter C and the RBF kernel parameter γ were adjusted based on the training set data using the grid search strategy in LibSVM.【Result】 The miR SVM had the sensitivity of 83.7% and specificity of 81.2% respectively on human data.The ROC of the model was plotted with the specificity and the sensitivity from the results,and gave an area under the ROC curve of 87.71%.Interestingly,the miR SVM classifier built on human data can correctly identify up to 89.2% of the real miRNAs from 20 other species.【Conclusion】 The successful detection of mature strand on the precursors provides a reliable method for predicting mature miRNAs from their precursors. |
| Key words: miRNA mature strand prediction Support Vector Machine |
