search

artic-network / fieldbioinformatics

The ARTIC field bioinformatics pipeline
MIT License
110 stars 69 forks source link

Muscle 5.1 breaks the excution of `artic minion` #105

Closed pdp10 closed 2 years ago

pdp10 commented 2 years ago

As you do not pin the version of your dependency package muscle to 3.8, the higher version 5.1 is now chosen by conda.

The 5.1 version has a different command set and your code now raises the following error: 

Running: muscle -in 20200311_1427_X1_FAK72834_a3787181_barcode07.muscle.in.fasta -out 20200311_1427_X1_FAK72834_a3787181_barcode07.muscle.out.fasta

Invalid command line
Unknown option in
thomasbtf commented 2 years ago

Having the same problem here. This also affects ARTIC wrappers such as ncov2019-artic-nf.

tavareshugo commented 2 years ago

Having same issue here using the ncov2019-artic-nf.

Although, from the environment.yml it does seem that the correct version is specified.

I think the problem might be with the conda recipe, which specifies version >=3.8.

So, I guess we need to open an issue/pull request in there instead?

thomasbtf commented 2 years ago

Did that: https://github.com/bioconda/bioconda-recipes/pull/33172

However, the tests didn't pass and I haven't had time to take a closer look.

If anyone can lend a hand, please feel free to do so!

tavareshugo commented 2 years ago

I'm not really experienced with conda-recipes, but I can't really understand what the problem is from the logs. From line 633, it seems like the installation of the environment fails (or was killed for some reason), which I guess is then triggering the error later when it tries to activate that environment and run the artic-specific tests?

My hack for the ncov2019-artic-nf in particular was to change the nanopore environment.yml to include - muscle=3.8 in the dependencies list.

BioWilko commented 2 years ago

Hi guys, we're hoping to bump the version to 1.2.2 for an urgent bugfix for the scheme downloader very soon; as soon as that's done I'll look at the conda recipe and pin muscle, I've had a quick look at the build fail logs and I haven't made much headway with it either but will have a proper look soon.

Edit: It looks like the free CircleCI node they use for testing is running out of RAM so I suspect I'll have to talk to someone at bioconda, will update here soon hopefully.

BioWilko commented 2 years ago

The recipe for v1.2.2 has now been merged on the bioconda recipes repo, it should take about 30 mins to propagate.