AVE cannot parse valid GFF file

[mkempenaar]

Firstly, because a proper GFF file should contain a version number in its header (like '##gff-version 3') which causes AVE to see this header as a chromosome ID.

Also, the validator that I use (an installable one, from http://genometools.org/) says that the attributes column cannot contain keys containing capitals since these are reserved for a number of keys thus failing for keys like 'Change', 'Strain', etc.

Furthermore, gene annotations are quoted strings that could contain semicolons which will crash ave_tools.

[mkuzak]

Hi Marcel,
I do agree for the possibility of commented line in gff file. I'll add this to the parser.

According to gff3 specification (http://www.sequenceontology.org/gff3.shtml) keys with capital letters are allowed.
"All attributes that begin with an uppercase letter are reserved for later use. Attributes that begin with a lowercase letter can be used freely by applications. "
I understand, that using upper case can in the future cause problems, in case attributes used by AVE will be added to gff specification. I decided to use upper case to stick to format used by 1001 genome project. I have to think what is better.

Can you give example of gene annotation with quoted string?

cheers,
Mateusz

[mkempenaar]

Hi Mateusz,

Thanks for the quick response. I've removed the gff version header and replaced the semicolons from the files and that worked just fine. See a few example lines of a gff file that fails: https://gist.github.com/mkempenaar/eb1b3d341e0b0b09cd1a

The full file can be downloaded from: http://solanaceae.plantbiology.msu.edu/data/PGSC_DM_v3_2.1.11_pseudomolecule_annotation.gff.zip

Thanks!

[mkuzak]

thank you for this example, I'll see what I can do, so that the import works on those quoted attribute values,

Mateusz