fasta2agp.pl error

[38293649]

Hi, I am using the latest version of funannotate to annotate a bird genome, but the final annotate step reported a error that I can't fix. This is the screenshot of that error, please help, thank you!

All the best,
De Chen

[nextgenusfs]

What characters do you have on the DNA sequence it is complaining about? I would always assume that only GATCN are acceptable for an assembly.

[waldeyr]

I have got the same error by running (Docker:latest) the following command:

funannotate annotate -i G1_predicted_genes/ --force --cpus 4 -o G1_annotated_genes/

OS: Debian GNU/Linux 10, 4 cores, ~ 16 GB RAM. Python: 3.7.9
Running 1.8.4
...
CMD ERROR: perl /venv/lib/python3.7/site-packages/funannotate/aux_scripts/fasta2agp.pl scaffolds.fa
(None, b"Will write contigs to file 'contigs.fsa' and AGP to STDOUT\nIllegal characters in sequence\nGATGTGAT...TG\n")

I'm not sure, but I suspect of the "\n" character at the end of the sequence.

[nextgenusfs]

Are either of you able to share your genome fasta file so I can take a look?

[38293649]

Thank you for your reply!
I think i know the reason. There are two scaffolds in the genome with all N, the fasta2agp.pl stopped when it processed one of them. I deleted these two scaffolds and only run fasta2agp.pl, it worked. However, if I run funannotate annotate -i fun --busco_db aves --sbt fun/template.sbt --cpus 40, the genome in the annotate_misc folder will include the two problematic scaffolds again and report error. How can I use the edited genome to run the funannotate annotate, thank you!

All the best

[nextgenusfs]

It is using the GBK and TBL files from the predict_results folder. So assuming there weren't any predictions on scaffolds that were all N's... than you should just be able to re-run your funannotate predict command with the properly edited genome FASTA file -- that way it will get incorporated into the output GBK file and then subsequently used in funannotate annotate.

[waldeyr]

Thank you for your reply!
I think i know the reason. There are two scaffolds in the genome with all N, the fasta2agp.pl stopped when it processed one of them. I deleted these two scaffolds and only run fasta2agp.pl, it worked. However, if I run funannotate annotate -i fun --busco_db aves --sbt fun/template.sbt --cpus 40, the genome in the annotate_misc folder will include the two problematic scaffolds again and report error. How can I use the edited genome to run the funannotate annotate, thank you!

All the best

Thanks for this tip. I gonna look at my scaffolds tomorrow to verify the existence of only 'Ns'. I will report it here later.

[ferninfm]

I have encountered this same problem again after updating to the newest version 1.8.4.

I manage to complete the funannotate annotate runs by modifying the perl code of fasta2agp.pl to use ALL IUPAC ambiguity codes. I don't know if that may cause upstream problems upon submission to NCBI. In principle all ambiguities should be accepted, or at least have been previously acceptable...

If anyone is in a hurry and for whatever reason has ambiguities other than Ns in their genomes, the solution proposed in issue #511 solves this problem.

That is changing line 76 in the fasta2agpl.pl script
from } elsif ( $substring_sequence =~ m/^[ACGTN]+$/i ) {
to } elsif ( $substring_sequence =~ m/^[ACGTRYSWKMBDHVN]+$/i ) {

Al the best

[jimie0311]

It is using the GBK and TBL files from the predict_results folder. So assuming there weren't any predictions on scaffolds that were all N's... than you should just be able to re-run your funannotate predict command with the properly edited genome FASTA file -- that way it will get incorporated into the output GBK file and then subsequently used in funannotate annotate.

Hi, could you please explain what is the meaning of no any prediction on scaffolds?
If one scaffold only include tRNA, that means this scaffold will be all N's?
I met the same problem when using funannotate annotate with the error of "AGP to STDOUT\nIllegal characters in sequence\n\n" I checked the tbl file in predict_results, there are few short scaffolds which don't have cds but only tRNA predicted. Should I move them to rerun funannotate predict?

[nextgenusfs]

The error doesn't have anything to do with gene predictions but rather the DNA sequences -- ie you apparently have non IUPAC characters in your genome sequence. (this means any character that is not in this string ACGTRYSWKMBDHVN). However, what version of funannotate are you running -- at some point I put this check in early during funannotate predict.

[jimie0311]

Hi Jon,
I am using v.1.8.7. I checked the fasta2agpl.pl, it is ACGTRYSWKMBDHVN