Funannote2 annotate only product = hypothetical protein

[OceaneMion]

Hi,

I am running into a problem with funannotate2.

I used the module funannotate-addons to obtain functionnal annotation with interproscan, eggnog, antismash and signalP, that I then give to funannotate2 with the following command:

funannotate2 annotate -f Gd_00045.fasta -g Gd_00045.gff3 -a emapper.annotations.txt antismash.annotations.txt iprscan.annotations.txt signalp6.annotations.txt --out . --busco-lineage leotiomycetes --cpus 75 --species "Pseudogymnoascus destructans" --strain "Gd_00045"

The script run perfectly without any error, but all the proteins seems to be annotated as hypothetical proteins I gave you a small output example here:

chr_1001 ANNEVO gene 124153 124460 . + . ID=chr_1001-g12;
chr_1001 ANNEVO mRNA 124153 124460 . + . ID=chr_1001-g12.t1;Parent=chr_1001-g12;product=hypothetical protein;Note=Glutathione-dependent formaldehyde-activating enzyme,COG:S,EggNog:ENOG50COG3791;
chr_1001 ANNEVO exon 124153 124426 . + . ID=chr_1001-g12.t1.exon1;Parent=chr_1001-g12.t1;
chr_1001 ANNEVO exon 124447 124460 . + . ID=chr_1001-g12.t1.exon2;Parent=chr_1001-g12.t1;
chr_1001 ANNEVO CDS 124153 124426 . + 0 ID=chr_1001-g12.t1.cds;Parent=chr_1001-g12.t1;
chr_1001 ANNEVO CDS 124447 124460 . + 2 ID=chr_1001-g12.t1.cds;Parent=chr_1001-g12.t1;

I am not sure if it is a problem from funannote2 annotate (or intended and then why so?), but I used ANNEVO for gene prediction and obtained a gff3, but it does not seems to be the problem here.

So any help or idea why this occurs would be welcome, and thanks again for developping and updating this wonderful tool !

[rzelle-lallemand]

Hi @OceaneMion, sorry to go off-track, but can you share how you installed/configured interproscan and eggnog-mapper to be used with funannotate2-addons? I'm interested in migrating from funannotate to funannotate2, but I couldn't find any documentation about how to install the dependencies so that the funannotate2-addons "f2a" wrapper can find them.

See also nextgenusfs/funannotate2-addons#6

[OceaneMion]

Hi @rzelle-lallemand , no problem from memory:

First I installed pixi:
curl -fsSL https://pixi.sh/install.sh | bash

Clone and installed env:
git clone https://github.com/nextgenusfs/funannotate2.git
cd funannotate2
pixi install --locked

Activate and create folder to install funannote2 database
pixi shell
mkdir -p funannotate2_db
export FUNANNOTATE2_DB=funannotate2_db
funannotate2 install -d all

Then I add eggnog and download the databases
pixi add eggnog-mapper
wget http://eggnog5.embl.de/download/emapperdb-5.0.2/eggnog.db.gz
wget http://eggnog5.embl.de/download/emapperdb-5.0.2/eggnog.taxa.tar.gz
wget http://eggnog5.embl.de/download/emapperdb-5.0.2/eggnog_proteins.dmnd.gz

You can then uncompress them and put in a folder and export the path

Then I run with following command
f2a emapper -f proteins.faa -o . --emapper_path usr/funannotate2/.pixi/envs/default/bin/emapper.py --data_dir usr/funannotate2/.pixi/envs/default/lib/python3.11/site-packages/data --cpus 75

And for interproscan I run to install:
wget https://ftp.ebi.ac.uk/pub/software/unix/iprscan/5/5.78-109.0/interproscan-5.78-109.0-64-bit.tar.gz
wget https://ftp.ebi.ac.uk/pub/software/unix/iprscan/5/5.78-109.0/interproscan-5.78-109.0-64-bit.tar.gz.md5
md5sum -c interproscan-5.78-109.0-64-bit.tar.gz.md5
tar -pxvzf interproscan-5.78-109.0-*-bit.tar.gz

cd usr/funannotate2/
pixi shell
pixi add "openjdk>=11,<12"

Then run:
f2a iprscan -f proteins.faa -o interproscan --iprscan_path usr/02-GENE_ANNOTATION/02-Functionnal_annotation/interproscan/software/interproscan-5.78-109.0/interproscan.sh --cpus 75 --json

Hope it can help you, I am a bit stuck on how to interpret my data though as almost all of my annotation of genes in my gff3 seems to be written as product = hypothetical protein even though I have evidence with the output of interproscan, eggnog, antismash, and signalP. I am wondering is it also the case for you with funnanotate v1 ?

[nextgenusfs]

Hi @OceaneMion. This is expected behavior. It is due to NCBI rules on product definitions, effectively you can only have a product definition (description) if you have a common name also attached to a gene. I need to check in our current f2 annotate how we are dealing with this when external annotations are brought in from other sources, while on the surface it would seem obvious to then be able to describe more functional annotations, it isn't easy to infer (correct) gene names and product definitions. Mostly I'm being conservative here so that annotations will pass NCBI filters. Granted, the submission "rules" I grew up with may have changed in the last several years, so there is perhaps some room for improvement here. It also naturally feels like this is something an LLM could do quite well. But it doesn't mean it makes it that much easier to integrate into a general genome annotation tool.

Do you have any gene models that have common names (name= field)? If you do, they should have product= annotations as well.

[OceaneMion]

Hi @nextgenusfs,

Thanks a lot for your awnser, here is an exemple of a gene model with a name, but with product still written as hypothetical (if I understood correctly what you were asking) ?

chr_1001 GFFtk gene 1818720 1820094 . - . ID=FUN2_007574;Name=HUT1;
chr_1001 GFFtk mRNA 1818720 1820094 . - . ID=FUN2_007574-T1;Parent=FUN2_007574;product=hypothetical protein;Dbxref=PFAM:PF00892.27,PFAM:PF04142.22,PFAM:PF08449.18;Note=BUSCO:73604at147548 [leotiomycetes_odb12.2],UAA transporter family,COG:P,EggNog:ENOG50COG0697,InterPro:IPR013657 HUT1,InterPro:IPR037185 EmrE-like,PF08449 UAA,mobidb-lite disorder_prediction,PTHR10778 SOLUTE CARRIER FAMILY 35 MEMBER B,SSF103481 Multidrug resistance efflux transporter EmrE,Pathway:Reactome:R-MMU-174362 Transport and metabolism of PAPS,Pathway:Reactome:R-MMU-727802 Transport of nucleotide sugars,Pathway:Reactome:R-HSA-174362 Transport and metabolism of PAPS,Pathway:Reactome:R-HSA-727802 Transport of nucleotide sugars,Pathway:Reactome:R-DME-727802 Transport of nucleotide sugars,Pathway:Reactome:R-DME-174362 Transport and metabolism of PAPS,Pathway:Reactome:R-DRE-727802 Transport of nucleotide sugars,Pathway:Reactome:R-DDI-727802 Transport of nucleotide sugars,Pathway:Reactome:R-DDI-174362 Transport and metabolism of PAPS,Pathway:Reactome:R-CEL-727802 Transport of nucleotide sugars,Pathway:Reactome:R-CEL-174362 Transport and metabolism of PAPS,Pathway:Reactome:R-SCE-727802 Transport of nucleotide sugars,Pathway:Reactome:R-SPO-727802 Transport of nucleotide sugars,Pathway:Reactome:R-DRE-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-HSA-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-CFA-727802 Transport of nucleotide sugars,Pathway:Reactome:R-HSA-5619072 Defective SLC35A2 causes congenital disorder of glycosylation 2M (CDG2M),Pathway:Reactome:R-RNO-727802 Transport of nucleotide sugars,Pathway:Reactome:R-RNO-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-BTA-727802 Transport of nucleotide sugars,Pathway:Reactome:R-SPO-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-HSA-5663020 Defective SLC35A1 causes congenital disorder of glycosylation 2F (CDG2F),Pathway:Reactome:R-CEL-4085001 Sialic acid metabolism,Pathway:Reactome:R-HSA-5223345 Miscellaneous transport and binding events,Pathway:Reactome:R-MMU-4085001 Sialic acid metabolism,Pathway:Reactome:R-CEL-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-MMU-5223345 Miscellaneous transport and binding events,Pathway:Reactome:R-SPO-6787639 GDP-fucose biosynthesis,Pathway:Reactome:R-DME-6787639 GDP-fucose biosynthesis,Pathway:Reactome:R-HSA-4085001 Sialic acid metabolism,Pathway:Reactome:R-MMU-9939291 Matriglycan biosynthesis on DAG1,Pathway:Reactome:R-HSA-9913143 Antimicrobial resistance,Pathway:Reactome:R-SPO-4085001 Sialic acid metabolism,Pathway:Reactome:R-HSA-5619037 Defective SLC35A1 causes congenital disorder of glycosylation 2F (CDG2F),Pathway:Reactome:R-BTA-5223345 Miscellaneous transport and binding events,Pathway:Reactome:R-HSA-5619083 Defective SLC35A3 causes arthrogryposis,Pathway: mental retardation,Pathway: and seizures (AMRS),Pathway:Reactome:R-SSC-9939291 Matriglycan biosynthesis on DAG1;

[OceaneMion]

Hi @nextgenusfs,

I seems to still run into some problems with funannotate2.
When passing annotation files from external tools via -a/--annotations, some annotation_type values are not recognized by annotate.py, which only takes six exact keys (product, db_xref, name, note, ec_number, go_terms). Near-miss spellings produced by other tools, for example go_term instead of go_terms, or EC_number instead of ec_number seems to be parsed, merged, and reported in the annotation sources, but never transferred to the gene object. The run completes normally and the final summary.json reports 0 for those categories, with no warning. In my case this silently dropped 5,296 GO terms and 1,741 EC numbers.

As you can see here it is written as 0 for the output of signalP and interproscan:
"functional": {
"go_terms": 0,
"interproscan": 0,
"eggnog": 9449,
"pfam": 7426,
"cazyme": 265,
"merops": 232,
"busco": 3861,
"secretion": 0
}
}
}

Also I am not quite understanding how to use the --curated-names flag in regard to the --annotations (and where could I get those --curated-names data ?). I feel like in the help section the description are inverted for both. For exemple: --curated-names is described as Path to custom file with gene-specific annotations (tab-delimited: gene_id annotation_type annotation_value), but I feel like it is the description for --annotations no ?

Again thanks a lot for you help.