Cuffdiff FPKM question

weiw · 09-07-2013, 10:54 PM

Hello,

I am fairly new to the world of bioinformatics. I have data from an Illumina RNAseq run we did with Arabidopsis RNA. I ran the reads through the Cufflinks/Cuffdiff programs to get transcript levels for genes. I've noticed that for some genes, the program will group a few (2-4) genes/loci together and give them all the same FPKM. But if I search for these genes they are obviously separate genes and not isoforms of one gene. I believe it is because these genes are highly similar in sequence. But if I go and look at my aligned reads with a genome browser, many times most of the reads will only align to one of the genes. However in some instances both or all the genes show equal amounts of reads aligning. So my question is what exactly is the FPKM for these "grouped" genes? Is it for only one of the genes, or is it distributed amongst all the genes in the group? How do I find out what the FPKM for the individual genes are? Thanks in advance to anyone with any ideas!

nidhi.bioinfo · 10-18-2013, 04:21 AM

did you got any solution for your problem. I am facing the same issue.

weiw · 10-18-2013, 10:42 AM

Hi,

Yes I actually did. When I ran my Cuffdiff analysis, I used my merged.gtf file as the reference genome, not the original reference genome. My basic understanding is that the merged file takes into account your experimental data, as well the original reference genome, and creates a "merged" version of the two. So in the case of the multiple loci, the merged file decided based on my data that some reads span the entire region covering those two or more loci, thus the multiple loci with one FPKM. To fix this, I ran Cuffdiff again with the original reference genome, not the merged one. Hope that made sense...

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Solution found: For Cuffdiff/links 2.0.2 Make cufflinks FPKM match Cuffdiff FPKM	NGSfan	RNA Sequencing	4	04-16-2013 08:10 AM
Cufflinks FPKM is different than Cuffdiff FPKM !!	fuad193	RNA Sequencing	1	04-16-2013 06:22 AM
cufflinks FPKM >>> Cuffdiff FPKM	peromhc	Bioinformatics	6	10-17-2012 02:07 PM
Cufflinks and cuffdiff FPKM values	combiochem	Bioinformatics	12	10-14-2012 12:37 AM
FPKM calculated by Cuffdiff 1.3.0 VS 2.0.0	huijia	Bioinformatics	0	06-14-2012 01:41 PM

09-07-2013, 10:54 PM	#1
weiw Junior Member Location: Oregon Join Date: Sep 2013 Posts: 2	Cuffdiff FPKM question Hello, I am fairly new to the world of bioinformatics. I have data from an Illumina RNAseq run we did with Arabidopsis RNA. I ran the reads through the Cufflinks/Cuffdiff programs to get transcript levels for genes. I've noticed that for some genes, the program will group a few (2-4) genes/loci together and give them all the same FPKM. But if I search for these genes they are obviously separate genes and not isoforms of one gene. I believe it is because these genes are highly similar in sequence. But if I go and look at my aligned reads with a genome browser, many times most of the reads will only align to one of the genes. However in some instances both or all the genes show equal amounts of reads aligning. So my question is what exactly is the FPKM for these "grouped" genes? Is it for only one of the genes, or is it distributed amongst all the genes in the group? How do I find out what the FPKM for the individual genes are? Thanks in advance to anyone with any ideas!

10-18-2013, 04:21 AM	#2
nidhi.bioinfo Junior Member Location: new delhi Join Date: Oct 2013 Posts: 1	did you got any solution for your problem. I am facing the same issue.

10-18-2013, 10:42 AM	#3
weiw Junior Member Location: Oregon Join Date: Sep 2013 Posts: 2	Hi, Yes I actually did. When I ran my Cuffdiff analysis, I used my merged.gtf file as the reference genome, not the original reference genome. My basic understanding is that the merged file takes into account your experimental data, as well the original reference genome, and creates a "merged" version of the two. So in the case of the multiple loci, the merged file decided based on my data that some reads span the entire region covering those two or more loci, thus the multiple loci with one FPKM. To fix this, I ran Cuffdiff again with the original reference genome, not the merged one. Hope that made sense...