I would definitely like to add a test. Since there is already a test (but the data is missing) I was wondering if the data might still be somewhere. Any ideas? Otherwise I would take these: http://galaxyproject.github.io/training-material/topics/assembly/tutorials/general-introduction/tutorial.html

There is the possibility to input raw (and raw.gz) data. Any idea if there is a Galaxy data type for this?

No idea about the test data, sorry.
raw is very inexplicit do you know how this looks like?

seems so. at least velvet outputs this on startup. was wondering if we can parse this and use the value in the form. is there such a possibility?

In theory yes, in practice this function is deprecated :)

Looping in @Slugger70 here, he has used this tool a lot in the past.

For raw: I can only guess from a peek into the velvet sources that it is sequence data without any header info. I guess this is used only rareley. Also for raw no separate input of paired data seems possible. I guess we leave it out..?

@bernt-matthias I've found the test-data: http://lists.bx.psu.edu/pipermail/galaxy-dev/attachments/20100603/f9557958/attachment-0001.zip , see https://lists.galaxyproject.org/pipermail/galaxy-dev/2010-June/002760.html

I'd leave Raw out. I've never seen it used.. Bioconda Velvet is compiled with MAXKMERLENGTH=191, CATEGORIES=4, OPENMP=1, LONGSEQUENCES=1. You use the OPENMP settings by setting a couple of env vars. See the VelvetOptimiser wrapper for details.

        export OMP_NUM_THREADS=2 &&
        export OMP_THREAD_LIMIT=2 &&
        VelvetOptimiser.pl
            -t "\${GALAXY_SLOTS:-1}"
            -s $start_kmer
            -e $end_kmer
            -d out
            -f "

I couldn't figure out how to make the omp_num_threads set in a more dynamic way..

The velveth wrapper should now be functional.

• threading is enabled with export OMP_NUM_THREADS="\${GALAXY_SLOTS:-1}" &&. Unfortunately velvet uses one more thread than specified this way, but setting OMP_THREAD_LIMIT the same way seems to be dysfunctional in if the value is 1. Can we compute here? Then we could +1?
• I took a down sampled version of the GTN test data

Some questions on velvetg

• There was <option value="yes" selected="${input.metadata.paired_end_reads}">Yes</option> Does such a thing work? Linting fails with this.
• The test currently fails because the ftype="velvet" seems not to work. The data type seems odd to me: it consists of a (quite useless) html file and the Sequences and Roadmaps file in a sub directory. The velveth wrapper currently relies on the fact that the html file is seemingly autogenerated by the velvet data type class? (In the original version of this wrapper the velveth_wrapper.py script did this)
• I guess some of the conditions are unnecessary and can be replaced by simple bools?
• there is a TODO with some parameters which are currently not implemented in the wrapper. Should we include them? I guess they are related to the columbus extensions mentioned above.

@bernt-matthias I tried to enable testing for this repo.

@bernt-matthias You can compute using bash syntax, see e.g. https://github.com/galaxyproject/tools-iuc/blob/master/tools/trinity/trinity.xml#L11

Could/should this be moved to https://github.com/galaxyproject/tools-iuc?

Could: Yes. Should: I don't know.

Any comments on the question in my last comment? Then I could finish the work in this PR.

Moving to IUC: yes. But I think we can also do this after this PR.
I can not answer your questions, but @Slugger70 probably can. I'm wondering if velvet is worth to update nowadays.

I think its worth -- and if its only for the GTN tutorials.

closed via #564