Hello, I have a data set of micro-array experiments (single color illumina) that is strongly confounded by pedigree relationships. I am just interested in comparing patients and controls. I would like to give Leek and Storey's sva algorithm a go with limma, to help deal with confounding factors. Does anybody have some experience doing this? My main concern is that the sva algorithm assumes a linear model with an intercept and I would rather use one without it as it is most natural in order to simply compare two groups. Thanks, augusto