compare expression? cuffcompare or cuffdiff

[vebaev]

Hi All,
this forum is quite usefull for me I hope it will be in the future also

In general I have gtf files from 3 samples - wildtype and 2 more stressed (WT, A, B)) outputs from cufflinks.
I want to know stuff as how is one assembled sequence/gene difference expression trought these 3 points. Ans also top 100 altered genes from WT to A and from A to B etc

Can you suggest me what to use cuffcompare or cuffdiff?

PS - the organism genome is not sequenced!

thanks!

[vebaev]

anybody

[vebaev]

As I read more and more I figured out that I have to make the union file from my 3 samples giving the tree transcripts.GTF to Cuffcompare.

As I understood the cuffdiff takes the union GTF + Sam files (which comes from tophat alignments to ref.genome)

But what about the organism is not sequenced? there is no regerence genome...so how to run cuffdiff without sam-s.
Should I relay only to this union file from cuffcompare for the 3 sample compareson and forget cuffdiff?

[kmcarr]

vebaev,

You say that you ran cufflinks to produce GTF files; the input to cufflinks is a SAM file produced by TopHat. This is the SAM file you would also use as input to cuffdiff. How could you have run cufflinks if you did not align your reads to something and produce a SAM file as a result?

[vebaev]

thaks for reply,
the gtf was provided to me, and as I understood I have some BAM files with alignments, can I use them?

[kmcarr]

thaks for reply,
the gtf was provided to me, and as I understood I have some BAM files with alignments, can I use them?

BAM is the binary format of SAM. I don't recall at the moment if cuffdiff will accept a BAM file as input or if it must be in the SAM text format. If a SAM file is required you can easily generate a SAM version of a BAM file using the samtools command:

% samtools view -h -o <myOutPut.sam> <myInPut.bam>

Now as to whether or not you can use these files as input to cuffcompare and cuffdiff, I'm not sure. How were these files generated? I'm curious what the reads were aligned to to generate the BAM file if there is no reference genome as you say. Were the GTF files actually produced using the by cufflinks using the corresponding BAM files? If these things check out then you may be able to proceed with cuffcompare and cuffdiff.

[vebaev]

Thanks for the sam conversion idea.
I do not know where is coming from this BAM since it is made by company of my coleague maybe not with tophat. But afterwords it was used cufflinks for all there gtf-s