Hi,
I am trying to use cuffdiff to compare relative gene expression from some human cell line samples. I used the hg38 canonical reference genome with tophat for illimina to map the reads. I am now having trouble finding the appropriate GFF3/GTF file to use for cuffdiff. I tried the Ensemble GFF2 file, but it was not detected by the cuffdiff app.
Thanks,
Dan
Hello,
iGenomes has not released an annotation dataset for this genome yet:
http://support.illumina.com/sequencing/sequencing_software/igenome.html
There are specific attributes in an iGenomes reference that are utilized by Cuffdiff. These are often not included in other sources. You can read about the attributes needed for full functionality in the Cuffdiff manual and compare to what is included in reference files from other data providers.
The UCSC table browser will produce a reference GTF dataset for hg38, but it will be lacking these attributes, and in some cases the gene and transcript attributes will be the same value (triggering other issues with this tool package). However, it still may be the best choice if you wish to continue with hg38.
Your other option is to roll-back and use hg19 (start over from mapping) and incorporate the iGenomes GTF. Most people do this as far as I know.
Ensembl is not functioning most likely due to a chromosome identifier mismatch. You can confirm that is the problem for this dataset, plus any other reference annotation datasets you locate, by double checking for the issue using the help in this wiki. All inputs must be an exact match - meaning, based on the exact same reference genome, both sequence content and identifiers - or there will be problems.
http://wiki.galaxyproject.org/Support See section 2.11
Best, Jen, Galaxy team
