qpcrImpute {nondetects} | R Documentation |
This function models the missing data mechanism and uses an EM algorithm to impute the non-detect values in qPCR data.
qpcrImpute(object, dj=NULL, pyfit=NULL, groupVars=NULL, batch=NULL, tol=1, iterMax=100, outform=c("Single","Param","Multy"), vary_fit=TRUE, vary_model=TRUE, add_noise=TRUE, formula=NULL, numsam=5, linkglm = c("logit", "probit", "cloglog"))
object |
a qPCRset |
dj |
normalization values. If NULL, features with "control" in featureType(object) are used to normalize the data. If no control features are found, the data are not normalized. |
pyfit |
initial estimate of the relationship between the probability of a non-detect and average expression. If NULL, this relationship is estimated from the data. |
groupVars |
which columns in pData(object) should be used to determine replicate samples. If NULL, all columns are used. |
batch |
amatrix with control samples for each batch, if NULL, batch effect is not taken into account. |
tol |
likelihood convergence criterion of the EM algorithm. |
iterMax |
maximimum number of iterations of the EM algorithm. |
outform |
the form of the output requested.If "Single" performes a single imputation of missing values. If "Param" returnes estimated model parameters: mean and variance. If "Multy" performes a multiple imputation of missing values, and creats multiple data sets with imputed values. |
vary_fit |
if outform="Multy", includes the model uncertainty due to the logit of the probability of being missing. The default value is "TRUE". |
vary_model |
if outform="Multy", includes the model uncertainty due to the estimating mean of the data. The default value is "TRUE". |
add_noise |
if outform="Multy", introduses the variance component due to the random noise. The default value is "TRUE". |
formula |
specifies the model. |
numsam |
number of the datasets to be created if outform="Multy". The default value is 5. |
linkglm |
a link used for estimation of the missing data mechanism. |
The function returns a qPCRset object with non-detects replaced by their imputed values.
Valeriia Sherina
data(sagmb2011) tst <- qpcrImpute(sagmb2011, groupVars="sampleType", outform=c("Single"), batch=NULL, linkglm = c("logit"))