Where Do I Get A Gtf File With Proper Gene Ids For Read Counting?

Question

[Ryan Thompson]

I've just noticed that the knownGene.gtf file that you can download from the UCSC browser (for e.g. human) sets each transcript's geneID to be the same as the transcript ID. This causes problems when counting the number of reads mapping to each gene, since reads that align to shared exons of multi-isoform genes will be discarded. Is there any way to get these transcripts properly grouped into genes so that I can assign them geneIDs and count my reads properly?

Alternatively, is there some other human gene annotation I should be using for read counting?

Answer 1

[JC]

use Ensembl annotation, the genes have different IDs (ENSG\d+) than the transcripts (ENST\d+).

Answer 2

[Charles Warden]

I think you can download what you need from Illumina iGenomes:

http://cufflinks.cbcb.umd.edu/igenomes.html