基于深度网络投票的抗血管生成肽识别  

Identification of antiangiogenic peptides based on deep network voting

作　　者：李锦[1] 贺兴时[1] 梁芸芸 LI Jin;HE Xingshi;LIANG Yunyun(School of Science,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]西安工程大学理学院,西安710048

出　　处：《哈尔滨商业大学学报（自然科学版）》2024年第4期404-412,共9页Journal of Harbin University of Commerce：Natural Sciences Edition

基　　金：陕西省自然科学基金(2023-JC-YB-064)。

摘　　要：血管生成在各种疾病中,尤其是癌症的发病机制中起着关键作用,因此开发更加快速高效的抗血管生成肽(AAPs)智能识别工具尤为重要.基于多种特征工程、深度学习和集成学习构建了一个深度网络投票的识别模型iAAPs-DNV.采用AAindex编码、分组权重编码(EBGW)、K-间隔氨基酸对(KSAAP)、基于物理化学性质的二阶移动平均(SOMA)和BLOSUM62编码提取氨基酸序列的特征信息.利用软投票策略集成加入了注意力机制(attention)的双向长短期记忆网络(BiLSTM)和卷积神经网络(CNN),并通过全连接层输出识别结果.iAAPs-DNV模型在Main数据集和NT15数据集上的识别精度明显优于已有的识别模型,表明该模型能够高效准确地识别抗血管生成肽.Angiogenesis played a key role in the pathogenesis of various diseases,especially cancer,so the development of more rapid and efficient intelligent identification tools for anti-angiogenic peptides(AAPs)was particularly important.In this paper,a deep network voting identification model,iAAPs-DNV,was constructed based on multiple feature engineering,deep learning,and ensemble learning approaches.The feature information of amino acid sequences was extracted using AAindex coding,encoding based on grouped weights(EBGW),K-spacing amino acid pairs(KSAAP),second-order moving average(SOMA)derived from physicochemical properties,and BLOSUM62 coding.Subsequently,the soft voting strategy was employed to integrate the bidirectional long short-term memory network(BiLSTM)and the convolutional neural network(CNN),both of which incorporated the attention mechanism.Identification results were then outputted through a fully connected layer.The identification accuracies of iAAPs-DNV in the Main dataset and NT15 dataset were significantly superior to those of existing identification models,indicating that the model could efficiently and accurately identify AAPs.

关 键 词：抗血管生成肽 双向长短期记忆网络 卷积神经网络 软投票 注意力机制 

分 类 号：Q939.1[生物学—微生物学]