I have a question related to the use of cufflinks/cuffdiff usage: I have 2 Time Point paired-end samples (0' and 6h) with 3 technical replicates of different sample size due to sequencing.
This are the samples total reads:
My objective is to run tophat/cufflinks pipeline with normalization of samples
For the first paired-end replicates I ran the alignment as follows:
tophat -p 8 -G genes.gtf -o T0 genome.fa --library-type=fr-unstrand -g 10 -r 98 T0.rep1.r1 T0.rep1.r2
tophat -p 8 -G genes.gtf -o R6h genome.fa --library-type=fr-unstrand -g 10 -r 98 R6h.rep1.r1 R6h.rep1.r2
cufflinks -p 8 -o T0.rep1 T0.rep1/accepted_hits.bam -M mask --no-update-check
cufflinks -p 8 -o R6h.rep1 R6h.rep1/accepted_hits.bam -M mask --no-update-check
and for the other replicates the only parameter that I changed was the library-type to fr-secondstrand.
And my questions comes when analyzing mapped reads with cufflinks/cuffdiff.
1.- Is it possible to normalize between these replicates?
2.- If so when is recommended apply normalization (at cufflinks step or at cuffdiff?)?
3.- And which type of normalization do you recommend accordding to cufflinks/cuffdiff options?
In order to gain reads with technical replicates, I was thinking to pull down rep2 and rep3 into one sample per Time point. Do you recommend to do this step also?
Finally I paste below what would cuffmerge/cuffdiff commands look like:
cuffmerge -g genes.gtf -s genome.fa -p 8 assemblies.txt
cuffdiff -o diff_out -b genome.fa -p 8 -L T0,R6h -u merged_asm/merged.gtf T0.rep1/accepted_hits.bam,T0.rep2/accepted_hits.bam,T0.rep3/accepted_hits.bam R6h.rep1/accepted_hits.bam,R6h.rep2/accepted_hits.bam,R6h.rep3/accepted_hits.bam -T -M mask.gtf
Thanks for your help and support!!