I have RNA-Seq data and SNP data for 60 patients before (baseline) and one week after a drug treatment.
I'd like to use Matrix eQTL to identify eQTLs in these patients and see if there is a difference in the effect size of these eQTLs before and after treatment.
My first instinct was to run Matrix eQTL in all the samples together and include patient ID as a covariate, as samples from the same patient are likely to be correlated. Or is the error covariance parameter what should be used for this? The documentation states that this parameter can be set to numeric() for homoskedastic independent errors, but what about correlated errors?
Any guidance would be helpful.