Rattus norvegicus Rnor5.0 GTF annotation in GENCODE format ?

Question

[piotr.grabowski]

Dear BioStars,

I am using a piece of software (PARalyzer for PAR-clip data). The pipeline I am using requires a GENCODE format .gtf annotation. Since the official GENCODE rat annotation is not available, maybe somebody tried re-structuring the UCSC annotation to fit the GENCODE format ?

I already tried appending a second "filler" column to match the column order, but there are way more differences in the last multi-entry column. The home-made annotation with filler column didn't work (all works perfectly for GENCODE human and mouse annotation).

I would be very thankful for any tips or maybe files ;)

Have a nice weekend!

Piotr

Answer 1

[piotr.grabowski]

I tried using and playing around with Ensembl, UCSC and NCBI .gtf's with no success.

Few of the important differences are:

1. Names of chromosomes (GENCODE uses "chr1" instead of "1" or "chrM" instead of "MT"))
2. The second column in GENCODE format is the source of the annotation (ENSEMBL/HAVANA)
3. The 9th column (with key-value pairs) is quite different as well (e.g. non-GENCODE gtf doesn't contain "level" information).

It is possible to re-structure the .gtf from UCSC to such format and make it very similar, but if somebody already has something like this and knows it works then I would much appreciate using such tested .gtf then making my own and testing it while re-inventing the wheel...

Best,
Piotr

[Emily]

You want Ensembl. We make the GENCODE GTFs (for human and mouse) and we make the GTFs for all our other species in the same style.

[piotr.grabowski]

They are exactly in the same style ?

I used the newest rat Ensembl GTF, but had to make some minor changes - for example in human GENCODE gtf there is a "gene_type" not "gene_biotype" in the key:value column, exon numbers have no quotes in GENCODE GTF and the chromosome names are different (chr1 in GENCODE vs 1 in ENSEMBL).

I ended up writing a Python script in the end for making those minor changes. All seems to work now.

Best regards,
Piotr

[Tom]

Is this script available on GitHub? I am running into the same problem, and I also have the same problem of not wanting to spend time re-inventing the wheel. My pipeline works perfectly if I use GENCODE GTFs, but does not work if I use the Ensembl GTF for rat.

Answer 2

[Bert Overduin]

I have no idea what the difference between the GENCODE GTF format and regular GTF format exactly is, but have you had a look at the Ensembl GTF file for rat?