mglm(edgeR)
mglm()所属R语言包：edgeR
Fit Negative Binomial Generalized Linear Model to Multiple Response Vectors
符合负二项式的广义线性模型的多个响应向量
译者：生物统计家园网 机器人LoveR
描述----------Description----------
Fit the same log-link negative binomial or Poisson generalized linear model (GLM) to each row of a matrix of counts.
适合相同的log链接负二项分布计数矩阵的每一行或Poisson广义线性模型（GLM）。
用法----------Usage----------
mglmLS(y, design, dispersion=0, offset=0, start=NULL, tol=1e-5, maxit=50, trace=FALSE)
mglmOneGroup(y, dispersion=0, offset=0, maxit=50, trace=FALSE)
mglmOneWay(y, design=NULL, dispersion=0, offset=0, maxit=50, trace=FALSE)
mglmSimple(y, design, dispersion=0, offset=0, weights=NULL)
mglmLevenberg(y, design, dispersion=0, offset=0, start=NULL)
deviances.function(dispersion)
designAsFactor(design)
参数----------Arguments----------
参数：y
numeric matrix containing the negative binomial counts. Rows for tags and columns for libraries.
数字矩阵含有负二项式计数。行库的标签和列。
参数：design
numeric matrix giving the design matrix of the GLM. Assumed to be full column rank.
数字矩阵提供的GLM的设计矩阵。假定为列满秩。
参数：dispersion
numeric scalar or vector giving the dispersion parameter for each GLM. Can be a scalar giving one value for all tags, or a vector of length equal to the number of tags giving tag-wise dispersions.
数字标量或矢量的GLM为每个分散参数。可以给予所有标签的价值，是一个标量或向量的长度等于给标签明智的分散的标签的数量。
参数：offset
numeric vector or matrix giving the offset that is to be included in the log-linear model predictor. Can be a scalar, a vector of length equal to the number of libraries, or a matrix of the same size as y.
数值向量或矩阵给予抵消，是要在对数线性模型预测。可以是一个标量，向量的长度等于数字图书馆，或为y同样大小的矩阵。
参数：weights
numeric vector or matrix of non-negative quantitative weights. Can be a vector of length equal to the number of libraries, or a matrix of the same size as y.
非负的量化权重的数值向量或矩阵。可以是一个长度等于库的数量，或为y同样大小的矩阵向量。
参数：start
numeric matrix of starting values for the GLM coefficients. Number of rows should agree with y and number of columns should agree with design.
数字矩阵开始的GLM系数的值。行数应同意y和列数应与design同意。
参数：tol
numeric scalar giving the convergence tolerance.
数字标给人的收敛公差。
参数：maxit
scalar giving the maximum number of iterations for the Fisher scoring algorithm.
标给费舍尔得分算法迭代的最大数量。
参数：trace
logical, whether or not to information should be output at each iteration.
信息是否不逻辑，应该是在每一次迭代输出。
Details
详情----------Details----------
The functions mglmLS, mglmOneGroup and mglmSimple all fit negative binomial generalized linear models, with the same design matrix but possibly different dispersions, offsets and weights, to a series of response vectors. mglmLS and mglmOneGroup are vectorized in R for fast execution, while mglmSimple simply makes tagwise calls to glm.fit in the stats package. The functions are all low-level functions in that they operate on atomic objects such as matrices. They are used as work-horses by higher-level functions in the edgeR package.
职能mglmLS，mglmOneGroup和mglmSimple所有的拟合负二项分布广义线性模型，设计矩阵相同，但可能是不同的分散，偏移量和重量，到了一系列的响应向量。 mglmLS和mglmOneGroupR为快速执行，而矢量mglmSimpleglm.fit在统计包tagwise检测。的功能是所有低级别的功能，他们对原子对象，如矩阵操作。它们被用来作为工作马由磨边包在更高级别的功能。
mglmOneGroup fits the null model, with intercept term only, to each response vector. In other words, it treats the libraries as belonging to one group. It implements Fisher scoring with a score-statistic stopping criterion for each tag. Excellent starting values are available for the null model, so this function seldom has any problems with convergence. It is used by other edgeR functions to compute the overall abundance for each tag.
mglmOneGroup适合空模型，只有截距项，每个响应向量。换句话说，它把属于一组库。它实现了与费舍尔得分为每个标签的得分统计停止准则。良好的开端值可为空模型，所以这个功能很少有任何问题与收敛。其他磨边功能，它是用来计算每个标签的整体丰。
mglmLS fits an arbitrary log-linear model to each response vector. It implements a vectorized approximate scoring algorithm with a likelihood derivative stopping criterion for each tag. A simple line search strategy is used to ensure that the residual deviance is reduced at each iteration. This function is the work-horse of other edgeR functions such as glmFit and glmLRT.
mglmLS适合每个响应向量的任意数线性模型。可能衍生停止为每个标签的标准，它实现了一个矢量近似得分算法。用一个简单的线搜索策略，以确保在每次迭代中减少残留偏差。此功能是其他磨边机的功能，如glmFit和glmLRT马的工作。
mglmSimple is not vectorized, and simply makes tag-wise calls to glm.fit. This has the advantage that it accesses all the usual information generated by glm.fit. Unfortunately, glm.fit does not always converge, and the tag-wise fitting is relatively slow. mglmLevenberg is similar to mglmSimple, but makes tagwise calls to glmnb.fit in the statmod package instead of glm.fit. glmnb.fit implements a Levenberg-Marquardt modification of the scoring algorithm to prevent divergence.
mglmSimple没有量化，只是使标签明智的检测glm.fit。这样做的好处，它访问所有常用的信息生成glm.fit。不幸的是，glm.fit并不总是收敛，标记明智的拟合相对缓慢。 mglmLevenberg是mglmSimple类似，但glmnb.fit，在statmod包，而不是glm.fittagwise检测。 glmnb.fit实现得分算法的Levenberg马夸特的修改，以防止分歧。
All these functions treat the dispersion parameter of the negative binomial distribution as a known input.
所有这些功能，把分散的一个已知的输入负二项分布参数。
deviances.function simply chooses the appropriate deviance function to use given a scalar or vector of dispersion parameters. If the dispersion values are zero, then the Poisson deviance function is returned; if the dispersion values are positive, then the negative binomial deviance function is returned.
deviances.function只是选择适当的越轨行为的功能，使用色散参数的标量或向量。如果色散值是零，则返回泊松越轨功能，如果是积极的色散值，然后返回负二项分布偏差功能。
值----------Value----------
mglmOneGroup produces a vector of length equal to the number of tags/genes (number of rows of y) providing the single coefficent from the GLM fit for each tag/gene. This can be interpreted as a measure of the 'average expression' level of the tag/gene.
mglmOneGroup产生一个长度等于标签/基因提供的GLM适合每个标签/基因的单一系数（数量y行）向量。这可以解释为标记/基因的平均表达水平的措施。
mglmLS produces a list with the following components:
mglmLS产生以下组件的列表：
参数：coefficients
matrix of estimated coefficients for the linear models
线性模型的估计系数矩阵
参数：fitted.values
matrix of fitted values
矩阵拟合值
参数：fail
vector of indices of tags that fail the line search, in that the maximum number of step-halvings in exceeded
标签不合格的线搜索指数的向量，在，在超出的一步halvings的的最大数量
参数：not.converged
vector of indices of tags that exceed the iteration limit before satisying the convergence criterion
矢量标记指数超过前satisying收敛准则的迭代限制
mglmSimple produces a list with the following components:
mglmSimple产生以下组件的列表：
参数：coefficients
matrix of estimated coefficients for the linear models
线性模型的估计系数矩阵
参数：df.residual
vector of residual degrees of freedom for the linear models
向量的线性模型的残差自由度
参数：deviance
vector of deviances for the linear models
向量的线性模型deviances
参数：design
matrix giving the experimental design that was used for each of the linear models
矩阵给每个线性模型的实验设计
参数：offset
scalar, vector or matrix of offset values used for the linear models
用于线性模型的偏移值的标量，向量或矩阵
参数：dispersion
scalar or vector of the dispersion values used for the linear model fits
用于线性模型的色散值的标量或矢量适合
参数：weights
matrix of final weights for the observations from the linear model fits
从线性模式适合最终权重矩阵的意见
参数：fitted.values
matrix of fitted values
矩阵拟合值
deviances.function returns a function to calculate the deviance as appropriate for the given values of the dispersion.
deviances.function返回一个函数来计算适合于分散给定值的偏差。
designAsFactor returns a factor of length equal to nrow(design).
designAsFactor返回一个长度为等于nrow(design)的因素。
作者（S）----------Author(s)----------
Davis McCarthy, Yunshun Chen, Gordon Smyth
参见----------See Also----------
glmFit, for more object-orientated GLM modelling for DGE data.
glmFit，为更多的面向对象的胃排空数据的GLM模型。
举例----------Examples----------
y<-matrix(rnbinom(1000,mu=10,size=2),ncol=4)
dispersion <- 0.1
## Fit the NB GLM to the counts[＃适合计数的NB的GLM]
ave.expression <- mglmOneGroup(y, dispersion=dispersion)
head(ave.expression)
## Fit the NB GLM to the counts with a given design matrix[＃适合的NB的GLM计数与一个给定的设计矩阵]
f1<-factor(c(1,1,2,2))
f2<-factor(c(1,2,1,2))
x<-model.matrix(~f1+f2)
ave.expression <- mglmLS(y, x, dispersion=dispersion)
head(ave.expression$coef)
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-2-25 17:09:03
