Plan for supporting EggNOG v7 database?

[chengwbTOT]

Dear eggnog-mapper team,

I noticed that the EggNOG v7 data files (such as e7.proteins.dmnd, e7.og_info_kegg_go.tsv, etc.) have been released.

I would like to ask if there are plans to officially support EggNOG v7 in the download_eggnog_data.py script in the near future?

Additionally, could you please advise if it is currently possible to manually build the v7 database using create_dbs.py with the raw v7 files? I have downloaded the v7 datasets but I am unsure if the current database creation schema is compatible with the new v7 format.

Thank you for your hard work on this tool!

Best regards.

[diego00012138]

Is v6 or even v7 still not supported at present? v5 is based on data from 2019, which seriously affects the use of eggnog. This can also be seen from the citation volume of relevant literature

[chengwbTOT]

Hi, @diego00012138
Based on my recent testing, eggnog-mapper v2 still only supports the eggNOG v5 database. I tried running it with v6/v7 data, but it is not currently compatible.

I completely agree with your concerns. v5 is indeed becoming outdated given the rapid accumulation of genomic data since 2019, and the lack of support for newer versions is a significant limitation for current analyses. Hopefully, the developers will address this soon.

[fischuu]

I wrote for a MetaG pipeline we develop a quick+dirty annotator script for the new eggnog7 files, maybe it is helpful for one or another:

https://github.com/fischuu/eggnog7_annotator

It would need to be tested, and could get more functionality but for running annotations for existing protein sequences it should work. I am not associated at all to the EggNOG project, so this script is just build upon the published work.

[njohner]

@fischuu I appreciate the effort.
From what I saw by taking a very quick look at your pipeline (correct me if I'm wrong) is that you simply map against the database, while eggnog-mapper does much more: identifying the best hit in the orthogroup and using the precomputed phylogenies of that orthogroup to only transfer annotations from orthologs from that orthogroup and not from in-paralogs. It's hard to know how much difference that actually makes, but it's actually the feature I find interesting in eggnog-mapper. It's really sad that this great tool is not really maintained anymore.

[fischuu]

Yes, exactly, as I said it is was a quick and dirty approach to be able to include it into a bigger MetaG pipeline, so there are no fancy steps. A you described, I map against the database - and frankly, from comparing the hits between this and the V5 version, I do not see big difference. I hoped for more, but maybe that is because of the simplicity of the approach.

But I agree, it is really sad that the eggnog mapper is not maintained. As I use it only in computational pipelines, database V6 and V7 were not not really useful for me without a tool that can be used from command-line. That was the motication for the quick+dirty solution, to have something to look at.

[maartenciers]

I wonder if it would be possible to to append @fischuu method with databases found on https://eggnogdb.org/downloads/ mainly e7.trees.tsv.gz to be able to use the precomputed phylogeny and add that orthology information to the results. This way we have a more updated alternative to the eggnogg-mapper as the sequence similarity part can easily be replicated with mmseqs2. I'm not that surprised to not hear there is no big difference in hits between V5 and V7 only thing missing is knowing if its ortholog or paralog to transfer annotation with more confidence. What do you think?