I'm looking for some guidance with an RNAseq experiment with the following experimental design: 2 groups (infected and uninfected) and for each group we have 2 time points (pre- and post-treatment). I have 20 individuals divided up into those 2 groups (10 per group). And each of those individuals has data for each of the 2 time points.
I would like to do 4 comparisons for our data:
a) infected pre-treatment vs uninfected pre-treatment (independent samples)
b) infected post-treatment vs uninfected post-treatment (independent samples)
c) infected pre-treatment vs infected post-treatment (dependent, paired samples)
d) uninfected pre-treatment vs uninfected post-treatment (dependent, paired samples)
Comparisons 'a)' & 'b)' seem like a fairly straightforward cuffdiff run. I have a handle on how to approach those analyses. But for 'c)' & 'd)' I am not sure how to maintain the paired sets of samples per individual across each condition in such a way that maintains the pairing relationships correctly when comparing the 2 conditions.
Looking through the cuffdiff manual, I do see that there is an argument that can tell cuffdiff our input is timepoints (-T). But I think that only compares samples in the same sequential order listed as condition A vs their mate in condition B. I don't think that does any kind of smart aggregation across each pairing to arrive at an overall differential expression between the 2 timepoints.
Then I also see that there is an option to use a 'contrast file', in which I can define specific pairings of samples to compare between conditions A & B. That sounds like I could setup my paired timepoints to be compared to each other, but again it sounds like all this is doing is setting up specific pairs to generate results for. I don't think it does any special statistics that utilize all the sets of pairs to arrive at some aggregate result that overall compares my 2 timepoints.
Can anyone advise me on how to properly use cuffdiff to perform the analyses I mentioned at the top of this post? Specifically help with 'c)' & 'd)' would be most appreciated.
I would like to do 4 comparisons for our data:
a) infected pre-treatment vs uninfected pre-treatment (independent samples)
b) infected post-treatment vs uninfected post-treatment (independent samples)
c) infected pre-treatment vs infected post-treatment (dependent, paired samples)
d) uninfected pre-treatment vs uninfected post-treatment (dependent, paired samples)
Comparisons 'a)' & 'b)' seem like a fairly straightforward cuffdiff run. I have a handle on how to approach those analyses. But for 'c)' & 'd)' I am not sure how to maintain the paired sets of samples per individual across each condition in such a way that maintains the pairing relationships correctly when comparing the 2 conditions.
Looking through the cuffdiff manual, I do see that there is an argument that can tell cuffdiff our input is timepoints (-T). But I think that only compares samples in the same sequential order listed as condition A vs their mate in condition B. I don't think that does any kind of smart aggregation across each pairing to arrive at an overall differential expression between the 2 timepoints.
Then I also see that there is an option to use a 'contrast file', in which I can define specific pairings of samples to compare between conditions A & B. That sounds like I could setup my paired timepoints to be compared to each other, but again it sounds like all this is doing is setting up specific pairs to generate results for. I don't think it does any special statistics that utilize all the sets of pairs to arrive at some aggregate result that overall compares my 2 timepoints.
Can anyone advise me on how to properly use cuffdiff to perform the analyses I mentioned at the top of this post? Specifically help with 'c)' & 'd)' would be most appreciated.
