We should probably also adjust for sample source or batch effects in the model, similar to the way we also want to adjust for transcript length #17 .
Right now, all samples are from similar source but in later versions we may include other sources, e.g. the Tex project data.