摘要
目的尝试通过构建数学模型来确定蛋白质中与RNA相互作用的氨基酸位点。方法从蛋白质结构数据库(protein data bank,PDB)中收集532例蛋白质与RNA相互作用的数据,对每个氨基酸提取150个特征,并利用机器学习方法构建2个与RNA结合的蛋白质位点预测模型。结果经过在同一数据集上与其他模型比较,所构建的模型具有更好的性能。结论该预测模型的建立,为研究人员获得与RNA结合的蛋白质位点提供了较好的生物信息学支持。
Objective To develop mathematical models for detecting RNA-binding residues.Methods Totally 532 protein-RNA complexes from the Protein Data Bank were extracted,and 150 features were calculated for each interacting or non-interacting residue.Then,two RNA-binding site prediction models were developed using machine learning methods.Results Test results showed that the two models had better performance than previous ones on the same dataset.Conclusion This work provides a better bioinformatics support for searching for RNA-binding residues in protein sequence.
出处
《军事医学》
CAS
CSCD
北大核心
2012年第11期830-834,共5页
Military Medical Sciences
基金
国家973计划资助项目(2010CB912801)
国家自然科学基金资助项目(31071157)