combat {swamp}

ComBat algorithm to combine batches.

Description

Performs ComBat as described by Johnson et al.

Usage

combat(g, o.withbatch, batchcolumn = NULL, par.prior = T, prior.plots = T)

Arguments

g

the input data in form of a matrix with features as rows and samples as columns.

o.withbatch

the batch annotation as a factor vector or within a dataframe that contains additional biological co-variates. make sure that the order of annotation is the same as in g. rownames (o) must be identical to colnames (g). when submitting a data.frame o.withbatch it can contain only factors.

batchcolumn

Required. Specify the batch column number of a dataframe ; set to 1 for a vector. All columns have to be factors, no NAs allowed.

par.prior

if 'T' uses the parametric adjustments, if 'F' uses the nonparametric adjustments. if you are unsure what to use, try the parametric adjustments (they run faster) and check the plots to see if these priors are reasonable.

prior.plots

if 'T' will give prior plots with black as a kernal estimate of the empirical batch effect density and red as the parametric estimate.

Details

The R-code of the ComBat algorithm has been taken from the webpage jlab.byu.edu/ComBat and input and output were adopted to the swamp package. ComBat uses parametric and non-parametric empirical Bayes frameworks for adjusting data for batch effects. The method is robust to outliers and performs particularly well with small sample sizes. ComBat can handle only categorical batch variables in its current development stage. Biological covariates can be added to the model (also categorical).

Value

A numeric matrix which is the adjusted dataset.

Note

R coded algorithm directly from Johnson WE

Author(s)

Martin Lauss

References

Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics. 2007 Jan;8(1):118-27.

Examples

# data as a matrix
set.seed(100)
g<-matrix(nrow=1000,ncol=50,rnorm(1000*50),dimnames=list(paste("Feature",1:1000),
          paste("Sample",1:50)))
g[1:100,26:50]<-g[1:100,26:50]+1 # the first 100 features show
# higher values in the samples 26:50
# patient annotations as a data.frame, annotations should be numbers and factors
# but not characters.
# rownames have to be the same as colnames of the data matrix 
set.seed(200)
o<-data.frame(Factor1=factor(c(rep("A",25),rep("B",25))),
              Factor2=factor(rep(c("A","B"),25)),
              Factor3=factor(c(rep("X",15),rep("Y",20),rep("Z",15))),
              Numeric1=rnorm(50),
              row.names=colnames(g))

##unadjusted.data
res1<-prince(g,o,top=10)
prince.plot(res1)

##batch adjustment for Factor 3
com1<-combat(g,o$Factor3,batchcolumn=1)
##batch adjustment for Factor 3; with covariate
com2<-combat(g,o[,c("Factor2","Factor3")],batchcolumn=2)

##prince.plot
prince.plot(prince(com1,o,top=10)) 
prince.plot(prince(com2,o,top=10))