How to use combat in R??

[KYR]

I'm trying to use combat to correct for batch effect in our dataset. The sample csv files contains genes in rows, and samples in columns, and expression values are from different microarray. We have in total 6 batches, and the sif file contains 3 columns : arrays, samples, batch (as per the manual). R generates an error message when running combat.. I'm not sure what to do from here. What am I doing wrong??


Library(sva)
dat = read.csv("Combat_matrix_input.csv");
sif = read.csv("sif.csv");

modcombat = model.matrix(~1, data=dat)

newdata = ComBat(dat=dat, batch=sif$Batch, par.prior = TRUE, mod = modcombat)
Found 6 batches
Error in cbind(batchmod, mod) :
number of rows of matrices must match (see arg 2)

[blakeoft]

I think the problem is that your modcombat matrix doesn't have the right dimensions. When you run

modcombat = model.matrix(~1, data=dat)

you'll get an n by 1 matrix with each entry equal to 1, where n is the number of records or lines in the csv file. What you really want is an m by 1 matrix full of 1's, where m is the number of fields or columns of the csv file, assuming that you don't have any covariates in your data. If you do have covariates, just make a vector that describes the covariate and pass that in to the model.matrix function. Like this

modcombat = model.matrix(~ covariates)