converting GFF to GTF

[efoss]

I want to convert a GFF file from Flybase into a GTF file and I'm having a lot of trouble figuring out how to do this. Here is a description of the difference between GTF and GFF:

http://genome.ucsc.edu/FAQ/FAQformat.html#format4

Does anyone know of an easy way to convert GFF to GTF? It looks like there is a perl script to do this, but I was unable to install a module required for that script and also the script looks a bit complex and I saw comments claiming that it is not dependable. Does anyone know of an easy way to convert GFF to GTF?

[westerman]

I sent you private email but I suggest getting GTFs from Ensembl. ftp://ftp.ensembl.org/pub/current_gtf

Or if you are doing Illumina work then iGenomes is a good resource.

The GFF to GTF conversion is a pain.

[efoss]

I sent you private email but I suggest getting GTFs from Ensembl. ftp://ftp.ensembl.org/pub/current_gtf

Or if you are doing Illumina work then iGenomes is a good resource.

The GFF to GTF conversion is a pain.

Thanks very much. I see that they have a Drosophila GTF file on Ensemble. It's 5.25 rather than 5.41, but I assume that as long as it's 5 point something that it will be compatible with my other files.

Thanks again for the help.

Eric

[upendra_35]

Use the below perl script for coverting GFF to GTF

##!/usr/bin/perl

use strict;
use warnings;
use Data:umper;

use File::Basename;
use Bio::FeatureIO;

my $inFile = shift;
my ($name, $path, $suffix) = fileparse($inFile, qr/\.gff/);
my $outFile = $path . $name . ".gtf";

my $inGFF = Bio::FeatureIO->new( '-file' => "$inFile",
'-format' => 'GFF',
'-version' => 3 );
my $outGTF = Bio::FeatureIO->new( '-file' => ">$outFile",
'-format' => 'GFF',
'-version' => 2.5);

while (my $feature = $inGFF->next_feature() ) {

$outGTF->write_feature($feature);

}

[kmcarr]

Use the below perl script for coverting GFF to GTF

##!/usr/bin/perl
 
use strict;
use warnings;
use Data::Dumper;
 
use File::Basename;
use Bio::FeatureIO;
 
my $inFile = shift;
my ($name, $path, $suffix) = fileparse($inFile, qr/\.gff/);
my $outFile = $path . $name . ".gtf";
 
my $inGFF = Bio::FeatureIO->new( '-file' => "$inFile",
 '-format' => 'GFF',
 '-version' => 3 );
my $outGTF = Bio::FeatureIO->new( '-file' => ">$outFile",
 '-format' => 'GFF',
 '-version' => 2.5);
 
while (my $feature = $inGFF->next_feature() ) {
 
$outGTF->write_feature($feature);
 
}

While it looks good on paper, in my experience this has rarely worked. Almost alway due to some non standard formatting of the attributes column (column #9) of the input GFF. But, as always, YMMV.

[efoss]

Use the below perl script for coverting GFF to GTF

##!/usr/bin/perl

use strict;
use warnings;
use Data:umper;

use File::Basename;
use Bio::FeatureIO;

my $inFile = shift;
my ($name, $path, $suffix) = fileparse($inFile, qr/\.gff/);
my $outFile = $path . $name . ".gtf";

my $inGFF = Bio::FeatureIO->new( '-file' => "$inFile",
'-format' => 'GFF',
'-version' => 3 );
my $outGTF = Bio::FeatureIO->new( '-file' => ">$outFile",
'-format' => 'GFF',
'-version' => 2.5);

while (my $feature = $inGFF->next_feature() ) {

$outGTF->write_feature($feature);

}

Hi Upendra,

Thanks for the suggestion. I just tried to run this and got this error message:


------------- EXCEPTION -------------
MSG: don't know what do do with directive: '##species'
STACK Bio::FeatureIO::gff::_handle_directive /Library/Perl/5.10.0/Bio/FeatureIO/gff.pm:537
STACK Bio::FeatureIO::gff::_initialize /Library/Perl/5.10.0/Bio/FeatureIO/gff.pm:115
STACK Bio::FeatureIO::new /Library/Perl/5.10.0/Bio/FeatureIO.pm:277
STACK Bio::FeatureIO::new /Library/Perl/5.10.0/Bio/FeatureIO.pm:297
STACK toplevel /Users/efoss/sequencing/Aida/RNAseq/test_GFF_to_GTF_script_102611_3.pl:16
-------------------------------------

I imagine that this is related to kmcarr's comment. I find it a bit weird that it is so hard to convert from GFF to GTF, since the GTF format is only changing the 9th column of the GFF format, and it seems like (if you have all the necessary information in the GFF file) that it should be trivial to convert it into the GTF format. Is all the information necessary for the GTF format necessarily included in the GFF format?

Eric
Seattle

[Simon Anders]

I imagine that this is related to kmcarr's comment. I find it a bit weird that it is so hard to convert from GFF to GTF, since the GTF format is only changing the 9th column of the GFF format, and it seems like (if you have all the necessary information in the GFF file) that it should be trivial to convert it into the GTF format. Is all the information necessary for the GTF format necessarily included in the GFF format?

The whole point of the GTF format was to standardise certain aspects that are left open in GFF. Hence, there are many different valid ways to encode the same information in a valid GFF format, and any parser or converter needs to be written specifically for the choices the author of the GFF file made. For example, a GTF file requires the gene ID attribute to be called "gene_id", while in GFF files, it may be "ID", "Gene", something different, or completely missing. Hence, a general GFF-to-GTF converter (as opposed to one converting only GFF files from a very specific source) needs to guess this from the data, which is non-trivial.

[efoss]

The whole point of the GTF format was to standardise certain aspects that are left open in GFF. Hence, there are many different valid ways to encode the same information in a valid GFF format, and any parser or converter needs to be written specifically for the choices the author of the GFF file made. For example, a GTF file requires the gene ID attribute to be called "gene_id", while in GFF files, it may be "ID", "Gene", something different, or completely missing. Hence, a general GFF-to-GTF converter (as opposed to one converting only GFF files from a very specific source) needs to guess this from the data, which is non-trivial.

Thank you. That's an excellent explanation.

Eric

[unique379]

Hello all,

I would like to have Genome coordinates of miRNA in .gtf format, is it possible to convert from .gff3 of miRBase ?? if yes then how ??


Thanks and looking forward for your kind reply