基于位点特异性打分矩阵的卷积神经网络预测SARS-CoV-2核衣壳蛋白的蛋白质二级结构

新型冠状病毒(SARS-CoV-2)有4种关键的结构蛋白,而核衣壳蛋白就是其中的1种.本实验从公开数据库NCBI上选取的SARS-CoV-2核衣壳蛋白质序列数据,分析SARS-CoV-2核衣壳蛋白与SARS-CoV核衣壳蛋白的序列相似性,对SARS-CoV-2核衣壳蛋白的理化性质和疏水性进行分析;在此基础上提出基于位点特异性打分矩阵的卷积神经网络,预测SARS-CoV-2核衣壳蛋白的8类蛋白质二级结构.研究结果表明,核衣壳蛋白的二级结构主要为无规卷曲,此结果可为抗病毒药物的研发与新型冠状病毒肺炎的诊断提供参考.

Prediction of the protein secondary structure of SARS-CoV-2 nucleocapsid protein through the convolutional neural network based on a position-specific scoring matrix

The novel coronavirus(SARS-CoV-2) has four key structural proteins, and the nucleocapsid protein is one of them. This research selects the sequence data of SARS-CoV-2 nucleocapsid protein from the public database NCBI, analyzes the sequence similarity between SARS-CoV-2 nucleocapsid protein and SARS-CoV nucleocapsid protein as well as the physicochemical properties and hydrophobicity of SARS-CoV-2 nucleocapsid protein. With this, it proposes a convolutional neural network with a position-specific scoring matrix to predict the secondary structure of eight proteins of the SARS-CoV-2 nucleocapsid protein. The results show that the secondary structure of this protein is dominated by random coils, which can shed some light on the research and development of antiviral drugs and the diagnosis of novel coronavirus pneumonia.