Question: my experimental design for cuffdiff
0
gravatar for F
3.2 years ago by
F3.4k
Iran
F3.4k wrote:

hi,

I have a paied-end RNA-seq study like below

Arabidopsis-heat for 2 hours

Arabidopsis-control for 2 hours

Arabidopsis-heat for 8 hours

Arabidopsis-control for 8 hours

Arabidopsis-heat for 24 hours

Arabidopsis-control for 24 hours

I am running cuffdiff like below

cuffdiff -o diff_out -b genome.fa -p 8 -u merged_asm/merged.gtf  204.bam, 205.bam, 206.bam, 207.bam, 208.bam, 209.bam

at first telling 

Warning: No conditions are replicated, switching to 'blind' dispersion method

then would heat treatment in time points be compared considering controls????  

thank you

 

rna-seq next-gen • 926 views
ADD COMMENTlink modified 3.2 years ago by andrew.j.skelton735.6k • written 3.2 years ago by F3.4k
1

cuffdiff doesn't really support very complicated designs. In your case it'll do pairwise comparisons between the samples (you don't have biological replicates?). If you need something more complicated then you'll need to use a different tool.

ADD REPLYlink written 3.2 years ago by Devon Ryan89k
1
gravatar for andrew.j.skelton73
3.2 years ago by
London
andrew.j.skelton735.6k wrote:

If I remember right, it's commas between replicates, spaces between phenotypes. So your command should be:

cuffdiff -o diff_out -b genome.fa -p 8 -u merged_asm/merged.gtf 204.bam,205.bam,206.bam  207.bam,208.bam,209.bam

If 204/5/6 are phenotype A, and 207/8/9 are phenotype B

edit: Devon was right, lesson learnt, always check the manual first. 

ADD COMMENTlink modified 3.2 years ago • written 3.2 years ago by andrew.j.skelton735.6k
2

Spaces between groups, commas and no spaces between replicates :) So:

cuffdiff -o diff_out -b genome.fa -p 8 -u merged_asm/merged.gtf 204.bam,205.bam,206.bam 207.bam,208.bam,209.bam

That makes parsing it a bit simpler.

ADD REPLYlink written 3.2 years ago by Devon Ryan89k

thank you both, I don;t have any biological replication only these six samples. then you mean I should better use something like deseq2?

ADD REPLYlink written 3.2 years ago by F3.4k
1

Without replicates you'll get questionable results regardless of what you use (I don't know if DESeq2 will even allow this...I suspect not). If you happen to want to use a model like ~treatment+time where time is a numeric value then you could still use DESeq2 or something similar, though you'd need to recreate the count values. If that's not the case (likely), then DESeq2 won't do anything for you.

ADD REPLYlink written 3.2 years ago by Devon Ryan89k

thank you for your explanations

ADD REPLYlink written 3.2 years ago by F3.4k

Ah damnit, I was on my phone and didn't check the manual, thanks for correcting me! 

ADD REPLYlink written 3.2 years ago by andrew.j.skelton735.6k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1111 users visited in the last hour