基于深度学习的PLK1 PBD活性预测

发布时间：2018-03-19 03:17

  本文选题：活性　切入点：深度信念网络　出处：《计算机应用研究》2017年01期 　论文类型：期刊论文

【摘要】：为了提高抗癌药物的发现效率并降低研发成本,针对基于PLK1此类结构和功能均高度保守的丝氨酸/苏氨酸蛋白激酶在多种肿瘤类型中高表达的特点,提出以PLK1 PBD为靶点的深度信念网络(deep believe network,DBN)抗癌活性研究方法。利用深度学习思想,对20 000个化合物的抗癌活性进行分析,并分别与ANN、SVM方法进行对比验证。实验结果表明,在同等条件下,DBN网络针对抗癌药物活性研究具有突出的优势,其平均预测活性的精确度可达91.05%,明显高于ANN和SVM,从而实现了对化合物抗癌活性的良好评估。
[Abstract]:In order to improve the efficiency of the discovery of anticancer drugs and reduce the cost of research and development, the high expression of serine / threonine protein kinase, which is highly conserved in structure and function of PLK1, was studied in various tumor types. The research method of anticancer activity of deep believe network (DBN) with PLK1 PBD as the target is proposed. The anticancer activity of 20 000 compounds is analyzed by using the idea of deep learning, and compared with the ANN PBD method. The experimental results show that, Under the same conditions, the research on anticancer drug activity of DDBN network has outstanding advantages, its average accuracy of predicting activity can reach 91.0555.It is obviously higher than that of ANN and SVM, thus realizing a good evaluation of the anticancer activity of compounds.
【作者单位】： 新疆大学软件学院;新疆大学网络中心;
【基金】：新疆维吾尔自治区研究生创新基金资助项目(XJGRI2015034)
【分类号】：R979.1;TP181
，

本文编号：1632587