gffread handle Mitochondrion gff3 has bugs?

[GuoqiLiu]

hello,
gffread handle animal Mitochondrion genomes has bugs? As follows:

gene4 gene=ATP8
MPQLDTST.LTTILSIFLALFIIFQLKISKYDFYHNPELTAKILKHNTP.ETK.TKIYLPLLLPL.
There are breakpoint in the middle . What is gffread support transl_table ?
Thanks !

[gpertea]

Indeed at this time gffread only supports the common, standard code for translations (transl_table 1).
As a workaround you could probably use the -x option instead, to output the CDSs and pipe the output into a mitochondrial-aware FASTA DNA-to-aminoacid translator, for example my seqmanip script (found in https://github.com/gpertea/gscripts, direct link to the raw file: https://raw.githubusercontent.com/gpertea/gscripts/master/seqmanip, it doesn't have any other dependencies). With seqmanip it would go like this:

gffread transcripts.gff -x- | seqmanip -T -B > proteins.fa 

(seqmanip has the -B option to enforce the mitochondrial translation table.)

[GuoqiLiu]

OK,I Know ! And whether I can use gffread to handle Bacteria genomes download from NCBI ?

[gpertea]

Not sure if I understand the question, if it's about supporting the non-standard translation tables for bacterial genomes (other transl_table values), I already mentioned that gffread does not (and neither does seqmanip).

If you are asking more generally about gffread handling annotation GTF/GFF3 data for bacterial genomes.. I haven't tested this but I suppose the answer is yes, it should work just the same (except for the translation issues that you already noticed). Not sure how the NCBI annotation looks like, for bacterial genomes -- if they always define genes with transcripts and/or exon/CDS features then it should work just the same. But if they use only "gene" features with no other sub-features, then one could use the new --gene2exon option of gffread which was specifically added to deal with some minimalistic GFF3 files which only have "gene" feature lines for single-transcript, single-exon genes.
But only use that option if needed.

[GuoqiLiu]

Thanks for your reply ,
Your understand is right ! I have used gffread to handle Bacteria genomes download from NCBI that can't come out error , so gffread can handle Bacteria.

[Maarten-vd-Sande]

Just dropping in to say that supporting multiple tables would be an awesome feature (for me). I haven't found any other tool that works as easy and fast as gffread.

Would you support a PR in this direction, if so, can you give some pointers? It's been over 5 years since I did anything serious with C++ so it might not result into anything useful though 😇 .. I can't seem to find where the table is defined..

[gpertea]

Sure, PRs are always welcome! As you can see, the actual codon table is in gclib/codons.cpp
I don't know where I got that code in gclib/codons.cpp from, seems unnecessary to waste 32K for a simple codon table like that (just to maximize the translation performance..). There are likely better ways to implement (multiple) codon tables..