Bahari Mohamad Hasan, Mahmoudi Mahmoud, Azemi Asad, Mirsalehi Mir Mojtaba, Khademi Morteza
Bioinformation. 2010 Jul 6;5(2):58-61. doi: 10.6026/97320630005058.
In this paper a new method based on artificial neural networks (ANN), is introduced for identifying pathogenic antibodies in Systemic Lupus Erythmatosus (SLE). dsDNA binding antibodies have been implicated in the pathogenesis of this autoimmune disease. In order to identify these dsDNA binding antibodies, the protein sequences of 42 dsDNA binding and 608 non-dsDNA binding antibodies were extracted from Kabat database and encoded using a physicochemical property of their amino acids namely Hydrophilicity. Encoded antibodies were used as the training patterns of a general regression neural network (GRNN). Simulation results show that the accuracy of proposed method in recognizing dsDNA binding antibodies is 83.2%. We have also investigated the roles of the light and heavy chains of anti-dsDNA antibodies in binding to DNA. Simulation results concur with the published experimental findings that in binding to DNA, the heavy chain of anti-dsDNA is more important than their light chain.
本文介绍了一种基于人工神经网络（ANN）的新方法，用于识别系统性红斑狼疮（SLE）中的致病性抗体。双链DNA结合抗体与这种自身免疫性疾病的发病机制有关。为了识别这些双链DNA结合抗体，从Kabat数据库中提取了42种双链DNA结合抗体和608种非双链DNA结合抗体的蛋白质序列，并利用其氨基酸的物理化学性质即亲水性进行编码。编码后的抗体被用作广义回归神经网络（GRNN）的训练模式。模拟结果表明，该方法识别双链DNA结合抗体的准确率为83.2%。我们还研究了抗双链DNA抗体的轻链和重链在与DNA结合中的作用。模拟结果与已发表的实验结果一致，即在与DNA结合时，抗双链DNA的重链比轻链更重要。
