I have performed ChIP-seq for multiple transcription factors on samples from multiple patients. All ChIPs for each patient where sequenced on different sequencing runs (i.e. TF1, TF2, TF3 for patient 1 was on one flowcell and  TF1, TF2, TF3 for patient 2 was on a separate flowcell etc). I am looking at how each transcription factor's binding affinity is changing in disease vs. normal. As such, I will be using DBA_REPLICATE as a blocking factor when I look at differential TF binding sites to account for the batch effects of each replicate being on a different flowcell. 

When dealing with a matched design for tissue samples (i.e. normal vs disease, with normal and disease being defined in the Tissue metadata column of the sample sheet), is there a need to set up a blocking factor if the category being used for dba.contrast() is the tissue type i.e normal vs disease?

Is it okay to use:

dba.contrast(x, categories = DBA_TISSUE, block = DBA_REPLICATE)

Or do you also need to include DBA_TISSUE as a blocking factor?

dba.contrast(x, categories = DBA_TISSUE, block = c(DBA_REPLICATE,DBA_TISSUE))

Thanks!

If you have normal and disease for each patient, and all the patient ChIPs (both normal and disease) were done in the same run, then the second one is correct.