摘要
由于microRNA在生物体系统中起着重要的调控功能,对microRNA进行快速有效的预测很有必要.本文通过使用蚁群算法和支持向量机相结合的思想,结合microRNA的前体pre-miRNA序列特征和结构特征,构造了一种microRNA的预测方法.通过采集Sanger和UCSE数据库中的人类阳性和部分阴性数据集进行学习和测试,同时使用J48和BP神经网络两种机器学习方法进行对比,实验结果显示,使用蚁群算法和支持向量机的方法预测pre-miRNA的识别率达97.471%,与另外两种方法相对比,识别率分别提高了8.736%和10.575%,预测的准确性有显著提高.
Since microRNA has important adjusting and controlling function in the organism system,it is crucial to predict it in a quick and effective way.Taking account of microRNA's precursor: pre-microRNA's sequences and structure characters,this paper puts forward a microRNA prediction method based on the combination of Ant Colony Algorithm and Support Vector Machine.Through learning and testing both of the known positive pre-microRNA database selected from Sanger center and the negative dataset extracted from Refseq sequences in human protein area from UCSE database,along with a comparison with the other two machine learning method of J48 and RBF neural networks,the experiment result shows that the accuracy of pre-microRNA prediction through the combination of Ant Colony Algorithm and Support Vector Machine is higher than RBF neural network and J48.Therefore,this prediction method can facilitate experimental identification of pre-microRNAs.
出处
《河北工业大学学报》
CAS
北大核心
2012年第1期5-8,共4页
Journal of Hebei University of Technology