基于氨基酸理化特征识别疾病相关的蛋白质与金属离子配体的结合位点  

Identification of Binding Sites of Disease-associated Proteins and Metal lon Ligands Based on Amino Acid Physicochemical Characterization

作　　者：邹向辉 冯永娥[1] ZOU Xianghui;FENG Yonge(College of Science,Inner Mongolia Agriculture University,Hohhot 010018,China)

机构地区：[1]内蒙古农业大学理学院,呼和浩特010018

出　　处：《内蒙古农业大学学报（自然科学版）》2024年第2期78-85,共8页Journal of Inner Mongolia Agricultural University(Natural Science Edition)

基　　金：国家自然科学基金项目(62262050);国家自然科学基金专项项目(62141204)。

摘　　要：蛋白质与金属离子配体的结合在维持蛋白质结构稳定和代谢控制等方面起重要作用。为了帮助研究者理解蛋白质与金属离子相互作用的分子机制,确定蛋白质与哪种金属离子配体的结合是非常必要的。目前大部分的研究只针对金属离子结合位点与非结合位点的预测研究,本文基于氨基酸理化特征结合机器学习算法构建模型,针对疾病相关的蛋白质与3种金属离子配体(Ca^(2+)、Mg^(2+)、Zn^(2+))的结合位点进行三分类的识别。首先,基于国际公共数据库资源,构建了疾病相关的蛋白质与3种金属离子配体的结合位点数据库。然后,在滑动窗口下,提取5种特征(PSSM,PAAC,PDC,HAAC,HDC),再结合2种机器学习算法对3种金属离子配体的结合位点进行识别。结果发现:在单特征预测中,使用位置特异性矩阵(PSSM)的预测结果最好,预测总精度(OA)达到72.6%。最后,做了特征融合,结果发现:其他特征在联合了位置特异性矩阵(PSSM)后,结果相较于其单特征,预测总精度均有较大提高。可见该模型对于疾病相关蛋白与金属离子配体的结合位点有较好的识别能力。The binding of proteins to metal ion ligands plays an important role in the formation,structural stability,metabolic control and disease production of their complexes.To help researchers understand the molecular mechanism of protein-metal ion interaction,it is necessary to determine which metal-ion ligands the proteins bind to.At present,most of the studies focus on the prediction of metal ion binding sites and non-binding sites.In this paper,the binding sites of disease-associated proteins to three metal ion ligands (Ca^(^(2+)),Mg^(^(2+)),Zn^(^(2+))) were identified.Firstly,a database of binding sites of disease-associated proteins with three metal ion ligands was constructed based on the international public database.Then,five features (PSSM,PAAC,PDC,HAAC,and HDC) were extracted and combined with two algorithms to predict the binding sites of three metal ion ligands.The results showed that among the single feature predictions,the best prediction results were obtained using position specificity matrix (PSSM) with an overall accuracy (OA) of 72.6%.Finally,among the feature fusion,the results showed that the prediction accuracies were improved greatly compared with single feature after joining the PSSM.It was shown that the model had good recognition ability for the binding sites of disease-associated proteins and metal ion ligands.

关 键 词：金属离子配体 位置特异性打分矩阵 亲疏水氨基酸组分 机器学习算法 

分 类 号：Q61[生物学—生物物理学]