Add genomes on server functionality

[elischberg]

No description provided.

[peterjc]

Can you point me at the Galaxy tool author documentation for this new (to me) functionality? I'm wondering e.g. if it requires a specific version of Galaxy (and if it would fail gracefully on an older one).

[wm75]

@peterjc https://docs.galaxyproject.org/en/latest/dev/schema.html#tool-inputs-param-options
not sure when this was added, but definitely long ago (before 2015 I'd say) :-)

[elischberg]

I had problems in handling the term "value_label" in the output name, because i didn't understand it quite well. Is it only for having "Protein" instead of "prot" in the name of the output?
I removed it for now, but I can insert it again, if I get to know how to use it.

Maybe an additional test for the genome on server functionality would be nice?

[peterjc]

Would the current nested conditionals break old workflows etc? Might therefore this be a safer approach:

     <param argument="-dbtype" type="select" display="radio" label="Molecule type of input">
        <option value="prot">protein from your history</option>
        <option value="nucl">nucleotide from your history</option>
        <option value="nucl-server">nucleotide from server</option>
      </param>

[wm75]

Nice :) Thanks for adding the new test!

You will need to bump the @VERSION_SUFFIX@ eventually.

[wm75]

Would the current nested conditionals break old workflows etc?

@peterjc in a sense, yes. Though of course only for people that are trying to run their workflow against a server that doesn't have the old tool version (requested by the workflow) installed. We normally consider this an acceptable drawback though I don't like making the life of users unnecessarily hard either.
In this case, however, there is no way to add the feature without this kind of breakage. Even your proposed simpler solution would require a new outside conditional to treat the new third selection differently, and again that would make workflow auto-updates to the new version impossible.

After testing a few alternatives I suggested to @elischberg to use the repeat elements (instead of the original multi-selects) because that lets you combine FASTAs from the history and server-provided ones when building the DB.
Was hoping to be able to propose a smaller change, but it didn't work out. Sorry for the extra reviewing effort and hope you find this solution acceptable.

[wm75]

btw, the use case behind the new feature is a viral primer scheme design tool that lets you exclude primer candidates based on them having matches in a custom BLAST DB, e.g. to avoid amplification of host DNA. With the new option, users will be able to create their desired subtraction DB as long as Galaxy Europe has the corresponding host reference installed.

[peterjc]

Given both @elischberg and @wm75 have been testing this (I have not), I'm happy in principle to merge this.

There are not any other imminent pull requests pending, so I suggest you bump the version and describe the changes in https://github.com/peterjc/galaxy_blast/blob/master/tools/ncbi_blast_plus/README.rst - something like:

2.14.1+galaxy1 - Use genome FASTA files on the server with ``makeblastdb`` (contribution from ...)

[peterjc]

Test failures in GitHub actions:

 "failure;ncbi_makeblastdb;0"
"failure;ncbi_makeblastdb;1"
"failure;ncbi_makeblastdb;2"
"failure;ncbi_makeblastdb;3"
"failure;ncbi_makeblastdb;4"

Looking at the HTML file tool_test_output.html via the All tool test results.zip available from the artefacts of the failed run:

Error: Unknown argument: "dbtypne"
Error:  (CArgException::eInvalidArg) Unknown argument: "dbtypne"

Happily that looks like an easy fix [already suggested by @wm75, which I have just committed. Hopefully won't complicate further updates to the pull request - ask if you need help with git]

[wm75]

Thanks @peterjc and @elischberg - this looks good to me then!

[peterjc]

Squash-and-merged, thank you both.

In theory this will auto-deploy to the Tool Shed... https://github.com/peterjc/galaxy_blast/actions/runs/8003981806

[peterjc]

Test Tool Shed failed - could be in flux?

Main Tool Shed also failed:

Could not update ncbi_blast_plus
Unexpected HTTP status code: 500: {"err_msg": "Metadata may have been defined for some items in revision 'b6893f57f8d8'.  Correct the following problems if necessary and reset metadata.<br/><b>ncbi_makeblastdb.xml<\/b> - This file refers to a file named <b>all_fasta.loc.sample<\/b>.  Upload a file named <b>all_fasta.loc.sample.sample<\/b> to the repository to correct this error.<br/>"}
Failed to update repository contents for repository 'ncbi_blast_plus' on the main Tool Shed.
Repository metadata updated successfully for repository 'ncbi_blast_plus' on the main Tool Shed.

Looks like we need to add all_fasta.loc.sample to https://github.com/peterjc/galaxy_blast/blob/master/tools/ncbi_blast_plus/.shed.yml

[peterjc]

Huh - the automated deployment did go though even without the all_fasta.loc.sample file, so there are actually two auto-deployments to the main Tool Shed.

Anyway, these changes are now live on the Tool Shed 🎉

[nsoranzo]

Test Tool Shed failed - could be in flux?

The Test TS has been down for the last 2 months.

[peterjc]

@nsoranzo thanks - I didn't double check the TTS status 👍

[wm75]

Anyway, these changes are now live on the Tool Shed 🎉

... meaning the new version should be available on Galaxy Europe by Monday.

Thanks a lot!