Entering edit mode
divyak
•
0
@divyak-23015
Last seen 4.7 years ago
Hi
I have an experiment with two different tissue samples taken from 8 patients. But not all 8 of them have both tissue samples. 5 have both tissue samples, 2 have only the first kind of sample and 1 has the second kind but not first. Additionally, one of the samples also has two technical replicates. How do I factor all this into DESEQ2 design?
- Can I do design = ~patient + sample, despite not all of my patients having both kinds of samples?
- Can I pick one out the two technical replicates alone for the analysis?
Thanks for your reply! What generally happens if DESeq2 is run without replicates, though?
Can you be more specific? I don't know what situation you are describing.
Sorry! In my experiment, I have 5 samples in each of the two groups (A and B) I am comparing. The 5 samples are from 5 patients. So they aren't technical replicates in that they aren't from the same library. We've taken a sample of A from a patient and then a sample of B from another patient and repeated this with 5 patients. Will DESeq2 still work then to do a Group A vs B comparison?
Yes you’d do patient + condition as the design.
Awesome! Thanks. But what if I have 8 on each group but only 5 match with patients? The other 3 are from different patients and do not have a match in group B. Should I discard those three or will patient + condition still work?
Yeah we can only make use of paired or not, whereas limma can strike a balance between these.
Thanks for your reply! What generally happens if DESeq2 is run without replicates, though?