Bad end parameter.

[fangbohao]

Hi there, I got the following error message when I ran 'agat_sp_filter_incomplete_gene_coding_models' on and GFF3 which produced from 'agat_sp_merge_annotations.pl'. Do you have any solution to fix the GFF3 file that produced from 'agat_sp_merge_annotations.pl'?

------------- EXCEPTION -------------
MSG: Bad end parameter (3). End must be less than the total length of sequence (total=2)
STACK Bio::PrimarySeq::subseq /usr/local/lib/perl5/site_perl/Bio/PrimarySeq.pm:452
STACK Bio::Seq::subseq /usr/local/lib/perl5/site_perl/Bio/Seq.pm:633
STACK toplevel /usr/local/bin/agat_sp_filter_incomplete_gene_coding_models.pl:135

I attached this GFF3 file for your reference from this link: https://drive.google.com/file/d/1iTqn0mSbVyBwT5lsf_D7zdb8ok9LFc2k/view?usp=sharing
The reference genome could be found here: https://s3.amazonaws.com/genomeark/species/Haemorhous_mexicanus/bHaeMex1/assembly_curated/bHaeMex1.pri.cur.20220203.fasta.gz

Thanks!
Bohao

General (please complete the following information):

• AGAT-v1.0.0
• AGAT installation/use: singularity
• OS: computation cluster

[Juke34]

Hi, I don't think the problem is from agat_sp_merge_annotations.pl. It is more likely a problem in the underlying annotations. One of these annotations contain a feature with start/end location out of the fasta sequence file. Either the different annotation have not been done using the same fasta reference. In such case you should not merge the annotations.

[fangbohao]

Hi Juke, thank you for pointing out this issue!

I realize that for some annotation files, I started with a soft-masked reference genome, and however for the rest starting with an un-masked genome, despite the same species. This might be the issue and will try to rerun things all from soft-masked genomes.

Thanks!
Bohao

[Juke34]

Masked or not it should not change the boundaries of the annotated features, I mean out of the sequences. Could you try agat_sp_filter_incomplete_gene_coding_models on each file to see if it works. If it does then potentially agat_sp_merge_annotations.pl might introduce some errors.

[fangbohao]

Hi Jacques, great suggestion! by running each file separately, I found which file has an issue. It was not an issue of the masked genome, but an issue of a specific annotation program.
Thanks!
Bohao

[fangbohao]

Hi Jacques, could I have your help to check a GFF3 file that could not pass 'agat_sp_filter_incomplete_gene_coding_models.pl'? This GFF3 is the only one that could not work with 'agat_sp_filter_incomplete_gene_coding_models.pl'. The other files work very well. The error message is still "Bad end parameter".

I have checked the features in this GFF3 - none exceed the coordinates in the reference genome chromosomes. The GFF3 was converted from a BED file produced by TOGA program using 'agat_convert_bed2gff.pl'. The pipeline seems ok.

I attached the GFF3, BED file, and genome files below. Thank you very much for taking a look and pointing out the issue in this GFF3!

GFF3: https://drive.google.com/file/d/1xAlfLtkG-m3hjiB20QIT_moDNw770AyR/view?usp=sharing
BED (the original output from TOGA): https://drive.google.com/file/d/1xfr1-acc1IsuLmhzwziQoJIEFx9JQRv7/view?usp=sharing
Reference genome (1Gb): https://drive.google.com/file/d/16-QMCiT2IELr1nZDqBuNH8_XG3EyNb5e/view?usp=sharing

Thanks!
Bohao

[Juke34]

Could it be a similar case: https://www.biostars.org/p/9552902/

[Juke34]

Right the problem is that the CDS is < 3 and we need at least 6 nt to be checked (3nt for a start codon, 3nt for a stop_codon)
e.g:

SUPER_11	TOGA	exon	2357069	2357070	.	-	.	ID=exon39600;Parent=4030
SUPER_11	TOGA	CDS	2357069	2357070	.	-	0	ID=CDS39600;Parent=4030
SUPER_11	TOGA	five_prime_UTR	2357071	2357070	.	-	.	ID=five_prime_UTR4030;Parent=4030

I will update the agat_sp_filter_incomplete_gene_coding_models.pl script to fix the problem

[fangbohao]

Hi Juke, thank you very much for finding and solving this!
Bohao