Release schedule

[matthdsm]

Hi

We'd like to start implementing exomiser in our pipelines, but we noticed the latest release is from a year ago and a lot of work been done in the dev branch.

Is there a schedule for new releases and/or updates to the reference dataset?

Thanks
Matthias

[julesjacobsen]

Hi Matthias, yes we're developing a new release at the moment. It'll hopefully be done soon* and will include a new data release.

I would recommend you develop using the current release and data and then upgrade when the new one is done. There will be a few changes which will require your attention, but nothing fundamentally different from the current setup required to deploy, run or update the application as part of a pipeline.

• Valve Time

[julesjacobsen]

@matthdsm hopefully you've seen https://github.com/exomiser/Exomiser/releases/tag/15.0.0

[matthdsm]

Now I have 😄 Thanks for the heads up!

[matthdsm]

To make my life truly easy, would you be able to make a conda recipe with a wrapper for the CLI? That would enable us to then create a nextflow module over at nf-core and really use the tool in various pipelines

[julesjacobsen]

Possibly, but that depends on how much of a pain it's going to be for me to make your life easier... If you could link to how this is supposed to be done I'll have a look. We already have docker images and homebrew formulae (brew install exomiser/tap/exomiser) which might already make your life easier.

I'm not keen on adding a third thing to maintain on top of what there already is, if one of those things is already sufficient. We specifically made the docker image with a shell so that NextFlow could use it as it didn't work with the distroless image.

[matthdsm]

Thanks for the info! I can definitely work with the existing docker images.

Cheers
M

[julesjacobsen]

Great, let me know if you need help setting up the nf-core stuff. Getting the data sorted is usually the most painful step, but there isn't really any way around having to obtain the data at least once. It is possible for multiple Exomiser instances to concurrently access the same data (we do this for our benchmarking and discovery) as its all read only, so creating a common data directory and running from that would be a good starting point.

[matthdsm]

I'm already looking into hosting the full reference data on S3 so it's more easily accessible.
Would also be nice to have the smaller ref set referenced in #623 as test dataset for the nf-core module.

I'll ping you over at nf-core when I've got a nextflow module POC

[matthdsm]

Hi again @julesjacobsen.

I'm trying out the docker image, but I'm having some weird behaviour. Where can I find the Dockerfile used to build the image?
For example:

• I can't open a bash shell inside the image. This is most likely due to a default entrypoint.
• It would be handy is the exomiser alias is already configured and ready to go when staring the image.
• For use with nextflow, the image needs to have procps installed and available on $PATH. Not sure if this is the case.
• The image exits after each command, which will cause job failures within nextflow

Thanks
Matthias

[matthdsm]

Also, it seems that running exomiser -V/--version does not produce any output.

[julesjacobsen]

Yeah, I'm not getting any output for --version either. However, if you're running the non-bash version 15.0.0 like this:

$ docker run docker.io/exomiser/exomiser-cli:15.0.0 --help
Exomiser: Phenotype-driven variant prioritisation for rare Mendelian disorders.
Usage: exomiser [-hV] [COMMAND]
  -h, --help      Show this help message and exit.
  -V, --version   Print version information and exit.
Commands:
  analyse  Runs an Exomiser analysis on a single sample
  batch    Runs a batch of analyses. This will avoid the initial application
             startup time for each sample if running singly, will allow the JVM
             to fully optimise performance, and maximise the use of any caches
             which have been enabled. In cases where hundreds or thousands of
             samples need running, it is recommended to run this command on
             batches of samples, in parallel jobs on several independent nodes.
Documentation: https://exomiser.readthedocs.io/en/latest/
Source: https://github.com/exomiser/Exomiser

You're not supposed to be able to open a bash shell as this is the default distroless image. The image you want is 15.0.0-bash. There is a bash script which was needed to start things, which is odd and needs a closer look at.

docker run -it docker.io/exomiser/exomiser-cli:15.0.0-bash 
root@067cced9612c:/# source enable_exomiser.sh \
 bash enable_exomiser.sh \
 exomiser 
Exomiser: Phenotype-driven variant prioritisation for rare Mendelian disorders.
Usage: exomiser [-hV] [COMMAND]
  -h, --help      Show this help message and exit.
  -V, --version   Print version information and exit.
Commands:
  analyse  Runs an Exomiser analysis on a single sample
  batch    Runs a batch of analyses. This will avoid the initial application
             startup time for each sample if running singly, will allow the JVM
             to fully optimise performance, and maximise the use of any caches
             which have been enabled. In cases where hundreds or thousands of
             samples need running, it is recommended to run this command on
             batches of samples, in parallel jobs on several independent nodes.
Documentation: https://exomiser.readthedocs.io/en/latest/
Source: https://github.com/exomiser/Exomiser

Can you open a new ticket with links to the requirements for the docker container for use in Nextflow and I'll take a look. It won't be soon as I have a bunch of deadlines next week. You're always welcome to submit a PR...

[matthdsm]

I've identified most of the issues and I'm preparing a PR with fixes over at https://github.com/matthdsm/Exomiser/tree/fix/docker

[matthdsm]

#628

[julesjacobsen]

See commit: 2743573