Error from cgat gtf2gtf --method=genes-to-unique-chunks

Question

[PK]

Hi,

I installed the cgat tool via mamba. I want to create the unique chunks at gene level using

--method=genes-to-unique-chunks

so i copied the code from one of the biostars code.

cgat gtf2gtf --method=genes-to-unique-chunks -I /reference_annotation/GRCh38.86.gtf > output.gtf

But i'm getting this error

self.transcript_id = other.transcript_id File "pysam/libctabixproxies.pyx", line 638, in pysam.libctabixproxies.GTFProxy.__getattr__ KeyError: 'transcript_id'

I supplied the entire GTF file without any filtering. Do i have to filter the specifically before i supply to the cgat?

Answer 1

[i.sudbery]

Yes, cgat assumes that the input GTF is valid according to the following specification:

http://mblab.wustl.edu/GTF22.html

This means that the transcript_id field is a manditory attribute.

However, modern ensembl GTFs contain gene lines that do not have a transcript_id. The easiest way to filter these out would be either:

awk '$3!="gene" '  /reference_annotation/GRCh38.86.gtf  | cgat gtf2gtf --method=genes-to-unique-chunks -L output.log > output.gtf

or

grep "transcript_id" /reference_annotation/GRCh38.86.gtf  | cgat gtf2gtf --method=genes-to-unique-chunks -L output.log > output.gtf

[PK]

Thanks for your reply. It's working now