It pops up frequently, see for example: https://github.com/Gaius-Augustus/BRAKER/issues/920

Also might be useful to look at Tiberius, which is what I think more effort is being put into

(that group is involved w/ the EBP, so having a workflow would be very useful…)

I have something that is similar to the dev version of genomeannotator (https://github.com/nschan/nf-annotate) with a bit of a plant focus and support for long cdna reads (or short reads). But a lot of it is pasa and evm, which seem to do a good job, but are not extremely fast.. Braker3 seems popular but I haven't put time into trying it out or figuring out how exactly it works. I was mainly wondering about the state of this it seems like it would be an important pipeline in the nfcore lineup of things that can be put into pipelines :)

Braker3 can utilize RNA-Seq short reads, long reads, proteins, and ab initio callers, and can use BUSCO to help training. But it’s brittle at several steps, primarily to lack of error checking or (on some HPC) lack of network access to remote data, e.g. BUSCO. But, there are Docker/Singularity containers available, so it’s possible to use those and work around some of the problems.

I see, tiberius seems to be tailored towards animals? Have you tried using tsebra outside of braker? It seems to do something similar as EVM

Yes, I did a short read (Illumina) RNA-Seq + long read RNA-Seq (each run also had protein evidence). These currently need to be run separately in Braker3 and merged using TSEBRA, which is pretty fast.

TSEBRA is included in their Braker3 Docker image for what it’s worth, but I think it’s also available separately

And yes Tiberius is tailored towards mammalian genomes I believe, but can be trained for others which is what I think they are working towards. It currently has other limitations (no UTR or isoforms yet) but the big improvement is sensitivity on ab initio calls, which IIRC is on par w/ Braker3 including RNA-Seq + protein

Hi! I've been working on genome annotation quite a lot in my previous job, and I made a pipeline using: • Repeat Modeler + Repeat Masker for the masking • Breaker3 for the structural annotation per se • TSEBRA for the merging of annotation (if needed) • AGAT for the cleaning of the GFF (sometimes necessary for subsequent steps) and for the translation to a protein fasta file • Interproscan for the functional annotation • Busco for QC I could annotate big plant genomes files (tried up to 4Gb) and it was quite robust. I definitely agree that we should revamp this pipeline and give it a new chance to shine, especially as genome annotation is very important! I think translating Braker3 to Nextflow is a good idea, although it requires quite a lot of work! But first of all, we need to revamp globally the pipeline. Does someone wants to do it? I can do it otherwise, on my free time. Cheers!

To continue kicking the dead horse, should we, the parties interested in having an annotation pipeline in nf-core, maybe try to get together and figure out what that should look like and if / which of the existing pipelines could be used to expand on? @Usman's genepal looks very nice and seems to be pretty much ready to go into nfcore? I think it would be nice to have support for long read transcriptome sequencing, but that should be rather straight-forward to add?

Thank you @Niklas Schandry Yes, good idea. Let’s plan a catchup meeting. I am in New Zealand time zone? It is 8 PM here and 10 AM at your place. This time works for me. I am available on most days during the next week. Folks at PFR have been talking about adding support for long read transcriptome. I am keen to hear your ideas as well.

Yeah, around "now" would be fine for me, I can do Mondays, Thur or Fri next week. I have implemented long-read support in the pipeline I created, I use minimap2 + bambu, which seems to work fine, but maybe there are alternatives to bambu? I think it would just be an additional branch that goes into BRAKER. If we used fastplong for read trimming & QC instead of technology specific tools, that should also be sort of a single workflow for ONT or HiFi-Isoseq and should be a rather straight-forward addition?

Yes, let’s start with that. I can also seek feedback from my team before the meeting. Let’s do Thursday. I assume today is also Thursday at your place?

Yes, it is Thursday morning here

I am going to pencil in 8 PM my time and 10 AM your time for next Thursday. Thank you!

Sounds good, it probably makes sense to ask around for feedback

@oliviercoen would you also like to join?

Hi guys! yes I'd like to join if possible 🙂 what is your time zone @Niklas Schandry?

UTC+2, I think we are in the same time zone

Hum, actually I won't be able to attend the meeting next Thursday. Anyway I think both of you know much more about genome annotation that me, so I'll let you lead the decisions

BTW, do we have an answer from the core team about what's best to do with the existing genomeannotator pipeline?

I dont think we even formally inquired 🙂 but i assume that no one is very attached to the existing genomeannotator, since it didnt even get to the release stage

I mainly tagged your since you seemed interested in putting some effort into this

Actually I am, but I'm sure you'll be better than me at figuring out the overall design

Hi! Do you guys have decided what you want to implement for the pipeline? I have myself created a little genome annotation pipeline: https://github.com/OlivierCoen/genome_annotator Do you think we could sort of merge of different pipelines with the one existing?

Hi! I'm about to submit a first PR for a complete renewal of the pipeline. However, I noticed that there are some big files in its git history... Is there any way to delete them from the history at some point? Once the new version of the pipeline is on dev, is it feasible to have one of the core team members perform a `git filter-repo`on it? Thanks!

set the channel topic: This channel is archived, please go to #C0C6WPPDT0R

<!channel> we are archiving this channel/pipeline, and opening #C0C6WPPDT0R to replace it

archived the channel