I just started using Tophat-cufflink on Galaxy today. But the example I found was quite simple (2 samples comparison) and I got a bit confused. I have the ILLUMINA RNA-seq data from 3 different niches and within each niche there were 3 replicates. I did the de novo assembly using transABbyss as there were no reference genome.
I named my samples as follows
SCC1 SCC2 SCC3
SAC1 SAC2 SAC3
SDC1 SDC2 SDC3
I mapped the pair-end reads of each sample against the de novo assembly. Then I was not sure if i should use the data from each sample to do the cufflinks assembly or combine the data from the same niche then do the cufflinks assembly (but i combined the data within same niche anyway).
Next when it came to cuffmerge, I merged all the data across different niches. Later, I did cuffquant analysis using the output of cuffmerge and when it asked for replicates 1,2,3, I input the datasets from same niche as replicates 1,2,3, respectively. Now I have 3 cbx data file corresponding to each niche (I think so).
I am still waiting for the output of cuffdif, but I am quite uncertain about what I just did.
Any idea is welcome.