When analyzing bulk RNA-seq data of matched pre and post treatment patient samples, is it important or necessary to correct for patient ID as a batch effect to see differences at post treatment vs pre? I have been advised to do so but I don't want to remove variation in my data if it is necessary.
For batch effect analysis I have seen and debated on contradictory views but in your case I assume the samples were collected at different time as patient may reported or recover at different time with varies phenotypic information. In my opinion correction of batch effect is required and removal of unwanted variation is the ultimate key to find potential answer from your research problem.
Hi, I totally agree with Devender Arora
As I understood it, the idea behind blocking/removal of batch effects is to emphasize the differences between the treatments and alleviating those caused by baseline expression between the patients.
Your question sounds like the classical textbook problem for application of blocking
This appears to be a sample pairing problem rather than a batch normalization problem - I suggest that you use a paired statistical test to account for the matchings between your pre- and post-treatment samples (and no batch adjustment).
- Badges
- Science topic
- Similar topics
- Comment