Replicates in Tophat

Question

[Parham]

Hi, how do you map reads using TopHat when you have several replicates? Does it work like this:

tophat -p 8 -o output_dir \
  ./path_to_bowtieindex \
  ./path_to_replicate1 \
  ./path_to_replicate2 \
  ./path_to_replicate3

I am omitting most of parameters I use to keep it short.

Answer 1

[Devon Ryan]

Technical or biological replicates? What you showed will work for the former, but not the latter. For the latter, you need individual runs of tophat, since you need the output in different files.

[Parham]

For biological replicates! Then I wonder what do they mean in Trapnell C. et al 2012 by this:

tophat -p 8 -G genes.gtf -o C1_R1_thout genome C1_R1_1.fq C1_R1_2.fq

Are those technical replicates?

[Devon Ryan]

Yes, those are technical replicates (well, what you showed are just paired-end reads, but if you had supplied multiple pairs like file1_1.fq,file2_1.fq file1_2.fq,file2_2.fq then those would be technical replicates). Unless my memory is incorrect, tophat will take multiple input files and output them to the same BAM file. That's great for technical replicates or any other case where a sample is divided across multiple files (not uncommon), but won't work for biological replicates.

[Parham]

Ok I got it the other way around you said this way is for biological replicates!

[Parham]

Naaah I got confused! The way I asked the question in the beginning is for Biological replicates right? I am proofreading myself now.

[Devon Ryan]

Well, you just said replicates, so that could mean either type.

[Parham]

Then I wonder what should I do for Gage/Pathview analysis, when I have my accepted_hits_1.bam, accepted_hits_2.bam and accepted_hits_3.bam - all corresponding to one condition and three biological replicates? How do I merge them together?