Question: Reference Genome unable to be detected on cuffdiff/cufflinks & Empty CuffDiff output files
gravatar for yena.oh
2.6 years ago by
yena.oh70 wrote:


I am currently following the tophat->cufflinks->cuffmerge->cuffdiff pipeline. Everything up until cuffmerge worked out well. I have been trying to run cuffdiff between 4 conditions x 3 replicates/condition, using the the cuffmerge output as the input gtf. After having done so, all cuffdiff output files, including the sqlite file, showed up as empty. I have previously performed the same analysis with the same parameters and annotation file and it went perfectly well. Not sure where the problem lies, since the annotation file contains all the attributes needed (i.e. gene_id, gene_name, p_id, transcript_id, tss_id. Example:

Seqname Source  Feature Start   End Score   Strand  Frame   Attributes

chr1    unknown exon    3214482 3216968 .   -   .   gene_id "Xkr4"; gene_name "Xkr4"; p_id "P15119"; transcript_id "NM_001011874"; tss_id "TSS26513";

I am trying to re-run cuffdiff, but now under "Perform Bias Correction" -> "Reference Sequence Data" -> "Using Reference Genome" the drop down menu is absent and says "No options available."

Is anyone else experiencing the same problem or have a solution?

Thanks, Yena

ADD COMMENTlink modified 2.6 years ago by Jennifer Hillman Jackson25k • written 2.6 years ago by yena.oh70


Cuffdiff worked when run without biased correction. There seems to be a correlation between empty cuffdiff output files and using the "bias correction" option. Any ideas?

ADD REPLYlink written 2.6 years ago by yena.oh70
gravatar for Jennifer Hillman Jackson
2.6 years ago by
United States
Jennifer Hillman Jackson25k wrote:


A reference genome must be supplied when Bias Correction is used.

The list of acceptable genomes is presented on the tool form based on the "database" metadata assignment of the inputs. Assign this first (pencil icon -> Edit Attributes -> Reference genome/Build), then run the tool.

If using a Custom reference genome, the same can be done - convert it to a Custom Build (User -> Custom Builds) first, then assign the newly created build to the data.

Hope this helps! Jen, Galaxy team

ps: Wiki link that contains details for a slightly different but related issue (same solutions will apply):

ADD COMMENTlink written 2.6 years ago by Jennifer Hillman Jackson25k

Hi Jen,

Thanks for the info. Unfortunately, I am still quite confused with the results I am getting. Using the exact same parameters, gtp file from cuffmerge, and bias correction cuffdiff sometimes works and sometimes produces emtpy outputs. Out of 5 times I have tried, it worked once.

After it worked once, I ran it again to confirm, with the same conditions, and I got empty files once again.

Also, under "bias correction"-> "locally cached" usually the reference genome is detected automatically: "Mouse (Mus Musculus): mm10." The issue is that sometimes this file is detected, and sometimes it is not. And even if the reference genome file is detected and selected, cuffdiff produces empty outputs.... and again, sometimes it doesn't. Any ideas?

ADD REPLYlink modified 2.6 years ago • written 2.6 years ago by yena.oh70

I have tried selecting some of the cuffdiff files that are "empty" and edited attributes to select "auto-detect," and the files still remain as empty...

UPDATE: after editing attributes 2 more times, that seemed to solve the problem. Empty -> 517MB. Thanks Jen!

ADD REPLYlink modified 2.6 years ago • written 2.6 years ago by yena.oh70
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 169 users visited in the last hour