Tabix With Gff3

Question

[kevin.mohamed1992]

I'm very new in BioInformatics (VERY) - please someone help with this:

first I run this command:

bgzip mydata.gff3

Then this is created:

data.gff3.gz

Then I run:

tabix -p gff data.gff3.gz

ths is the error I get

[get_intv] the following line cannot be parsed and skipped: 
[ti_index_core] the indexes overlap or are out of bounds

I know there is nothing wrong with my data (it is used in my application elsewhere)

Can someone point me in the right direction to solve this?

[marcela.uliano]

Hey guys!

Sorry about the simplicity of the question! But I'm trying for ours to open a gff in artemis to check some gene models and I can not!! The fasta and sorted bam file work well, but I can not get the gff together at all!!

So, I tried to sort and index it with tabix, but its not working!

The sort with the line bellow seems to go fine

(grep ^"#" in.gff; grep -v ^"#" in.gff | sort -k1,1 -k4,4n) | bgzip > sorted.gff.gz

But then, When I try

tabix -p sorted.gff.gz

It does not work! It says:

[E::get_intv] failed to parse TBX_GENERIC, was wrong -p [type] used? The offending line was: "itr6_2569_ AUGUSTUS CDS 1416 1422 . + 0 ID=itr6_2569_pilon_pilon_pilon_pi.g1.t1.cds;Parent=itr6_2569_pilon_pilon_pilon_pi.g1.t1" Segmentation fault (core dumped)

My gff is a PASA update output:

file: # original

itr6_6049_ AUGUSTUS gene 18202 46612 . - . ID=itr6_6049_pi.g765;Name=itr6_6049_pi.g765.t1

itr6_6049_ AUGUSTUS mRNA 18202 46612 . - . ID=itr6_6049_pi.g765.t1;Parent=itr6_6049_pi.g765;Name=itr6_6049_ pi.g765.t1

itr6_6049_ AUGUSTUS exon 46565 46612 . - . ID=itr6_6049_pi.g765.t1.exon1;Parent=itr6_6049_pi.g765.t1

I would appreciate any help.. This is really annoying me because its seems simples, I know =(

Thank you!

[Pierre Lindenbaum]

this is a new question , not a reply. Open a new question please

[marcela.uliano]

Oh sorry, Pierre! I'll do that!

[cmdcolin]

You should use "tabix -p gff filename.gff.gz" not just "tabix -p filename.gff.gz"

Answer 1

[Pierre Lindenbaum]

check that your GFF file is:

• sorted on chrom/start
• doesn't contain a blank line.

[kevin.mohamed1992]

thanks, but now there is one error: [ti_index_core] the file out of order at line 19

this is line 19 : ctgA example SNP 1000 1000 0.987 . . ID=FakeSNP1;Name=FakeSNP;Note=This is a fake SNP that should appear at 1000 with length 1

[Pierre Lindenbaum]

your file is not sorted.

[kevin.mohamed1992]

alright - thanks, any direction you can give me with sorting?

[cmdcolin]

There's a note in the tabix documentation for sorting a gff (reproduced below, but I needed to separate the bgzip step separately)

(grep ^"#" in.gff3; grep -v ^"#" in.gff3 | sort -k1,1 -k4,4n) > out.sorted.gff3
bgzip out.sorted.gff3

[cmdcolin]

Another method to sort gff is to use "genometools" and run

gt gff3 -sortlines input.gff > output.gff

This command also has a useful -tidy option that can clean up and validate very messy GFFs. It will rename the ID column though, unless -retainids is used

Answer 2

[cmdcolin]

Another command I found useful to sort gff using awk

awk '$1 ~ /^#/ {print $0;next} {print $0 | "sort -t\"\t\" -k1,1 -k4,4n -k5,5n"}' file.gff > file.sorted.gff

This is similar to the method from http://www.htslib.org/doc/tabix.html but it properly sets the tab delimiter on the sort command and avoids subshells

[Pierre Lindenbaum]

avoids subshells

well, there is a system call....

Answer 3

[Juke34]

AGAT and GFF3sort can sort GFF/GTF files to get proper tabix compliant sorting. More information here