collapse_annotation.py cannot process the gtf file generated by gffread

[biozzq]

Dear all,

To prepare the gtf file used in rnaseqc, I first converted the gff file to gtf file using following command,

gffread-0.12.7.Linux_x86_64/gffread -T -o out.gtf input.gff, however it give me error when running collapse_annotation.py out.gtf collapse.gtf

Traceback (most recent call last):
  File "collapse_annotation.py", line 294, in <module>
    annotation = Annotation(args.transcript_gtf)
  File "collapse_annotation.py", line 89, in __init__
    attributes.pop('transcript_type'), g, start_pos, end_pos)
KeyError: 'transcript_type'

Based on above error message, I added gene_biotype and transcript_type information to the end of each line.
perl -e 'while(<>){chomp; print $_," gene_biotype \"protein_coding\"; transcript_biotype \"protein_coding\";\n"}' out.gtf >processed.gtf

Finally, when running collapse_annotation.py processed.gtf collapse.gtf, another error occured.

Traceback (most recent call last):
  File "collapse_annotation.py", line 294, in <module>
    annotation = Annotation(args.transcript_gtf)
  File "collapse_annotation.py", line 89, in __init__
    attributes.pop('transcript_type'), g, start_pos, end_pos)
UnboundLocalError: local variable 'g' referenced before assignment

I attached the processed.gtf here. How should this be handled?
processed.zip

Thank you in advance.
Best wishes,
Zheng zhuqing

[francois-a]

RNA-SeQC requires GTF in the format specified at https://www.gencodegenes.org/pages/data_format.html, with a gene > transcript > exon hierarchy in the feature type column (additional features like CDS etc are also supported). Your GTF is missing gene features, it only has transcripts and exonic features.

[KristinaGagalova]

Is there a tool to convert a gtf to the required format? I am also having issues with that.
Thank you in advance