What Kind Of Inputs Does Cuffdiff Take?

Question

[narges]

Hi,

I have run tophat over the RNA-seq data and I got the output files : accepted_hits.bam, insertions.bed, junctions.bed, deletions.bed, unmapped.bam.

Now to find the DE genes I want to run cuffdiff over the results. Is it enough if I use the accepted_hits.bam files + gtf file as the inputs for cuffdiff or I need to also use any other output files from tophat as well? For example do I need to have accepted_hits.bam + insertions.bed or junctions.bed for each samples?

Thank you in advance

Answer 1

[Istvan Albert]

From the manual:

Cuffdiff takes a GTF2/GFF3 file of transcripts as input, along with two or more SAM files containing the fragment alignments for two or more samples.

http://cufflinks.cbcb.umd.edu/manual.html#cuffdiff_input

Answer 2

[blair]

The BED output files produced by TopHat are for those interested in discovering new transcripts and splice-junctions. For a DE experiment using samples from an organism with a well annotated genome (like human or mouse) then your best bet is to just take your accepted_hits.bam from TopHat and input it right into Cuffdiff along with a reference GTF/GFF (the ones from iGenomes are nice because they extra annotations like TSS that are needed to get full alternative splicing info from your data). Hope that helps!

[biomoon]

Hello! Can you show me the command line of Cuffdiff with accepted_hits.bam and a genome reference GTF? What I want to get is RPKM or FPKM. Thank you a lot!