weightedCondLogLikDerDelta {edgeR} | R Documentation |
Weighted conditional log-likelihood parameterized in terms of delta (phi / (phi+1)
) for a given tag/gene - maximized to find the smoothed (moderated) estimate of the dispersion parameter
weightedCondLogLikDerDelta(y, delta, tag, prior.n=10, ntags=nrow(y[[1]]), der=0, doSum=FALSE)
y |
list with elements comprising the matrices of count data (or pseudocounts) for the different groups |
delta |
delta ( |
tag |
tag/gene at which the weighted conditional log-likelihood is evaluated |
prior.n |
smoothing paramter that indicates the weight to put on the common likelihood compared to the individual tag's likelihood; default |
ntags |
numeric scalar number of tags/genes in the dataset to be analysed |
der |
derivative, either 0 (the function), 1 (first derivative) or 2 (second derivative) |
doSum |
logical, whether to sum over samples or not (default |
This function computes the weighted conditional log-likelihood for a given tag, parameterized in terms of delta. The value of delta that maximizes the weighted conditional log-likelihood is converted back to the phi
scale, and this value is the estimate of the smoothed (moderated) dispersion parameter for that particular tag. The delta scale for convenience (delta is bounded between 0 and 1).
numeric scalar of function/derivative evaluated for the given tag/gene and delta
Mark Robinson, Davis McCarthy
counts<-matrix(rnbinom(20,size=1,mu=10),nrow=5) d<-DGEList(counts=counts,group=rep(1:2,each=2),lib.size=rep(c(1000:1001),2)) y<-splitIntoGroups(d) ll1<-weightedCondLogLikDerDelta(y,delta=0.5,tag=1,prior.n=10,der=0) ll2<-weightedCondLogLikDerDelta(y,delta=0.5,tag=1,prior.n=10,der=1)