离子通道的混合密度参数估计与状态判别  

Mixture Density Parameter Estimates and States′classificatiion for Ion Channels

作　　者：李彩霞[1] 黎培兴[2] 关永源[3] 方积乾[1] 

机构地区：[1]中山大学医学统计与流行病学系,广东广州510080 [2]中山大学数学系,广东广州510275 [3]中山大学药学院,广东广州510080

出　　处：《数理统计与管理》2005年第6期62-66,共5页Journal of Applied Statistics and Management

基　　金：中山大学青年教师启动基金资助(2000)

摘　　要：本文针对多电导水平离子通道的由多个正态分布加权组成的混合分布特点,用EM迭代算法对混合分布中的参数进行极大似然估计,并在此基础上,利用混合分布中最可能的成分判断通道状态,从而还原通道潜在信号,克服了离子通道分析软件PCLAMP中参数估计与状态还原的缺陷。The current amplitudes of multi -level ion -channel show a mixture distribution which probability density contains m normal density components. In this paper, the maximum likelihood estimates （MLE） were given for the parameters in the mixture distribution via EM algorithm. Based on the mixture distribution, the most possible compo- nent in the mixture distribution is used to classify the channel＇s states and then to restore the underlying ion -channel signal. These methods in the paper overcome the shortcomings on parameter estimations and classification for states in the software PCLAMP.

关 键 词：离子通道 混合算法 EM算法 极大似然估计 

分 类 号：Q-332[生物学]