We can help you then reviewing your tool before it gets uploaded to the ToolShed ...
hey there... I'm working on writing a data manager to manage pangolin's data assets. there are 3 of these - pangolin-data, constellations and pangolin-assignment - and they get updated on different schedules (although pangolin-data and pangolin-assignments try and stach in sync). so there are going to be 3 data tables. this is all for a single purpose though, so I'm wondering: make 3 DMs? or make 1 DM with options to update 3 tables? (i.e. either 1, 2 or all 3 tables)
another DM question - how can the existing contents of the data table be queried? its not in the data table that gets passed to the DM. this was less of a concern for me before, but now I'm working on a DM that potentially could download 100s of MBs of data, and avoiding re-downloading what is already there would be useful.
@pvanheus .. the best would be galaxyproject/galaxy#12941 .. unfortunately I did not find time to fix the tests .. and it will only be available from the nextr release
Ah, OK. Fancy.. Then I think
$__app__.tool_data_tables sounds reasonable
Hi everyone, I am working on adding GPU support for the local galaxy runner. Currently, the support is in the PR stage but once it is finished, we wanted to see how would GPU support affect the speed of research and how much the GPU-supported tools would be used.
To this end, I have two main questions:
To this end, I have two main questions:
- Which tools (that already have GPU support) are frequently used by researchers? (Currently, I know that racon was a popular one, any tool similar to this that could benefit researchers would be great).
- Which machine learning tools are popular among researchers and which domain researchers mainly use ML tools that are in Galaxy? (Asking this, because many ML tools can easily be GPU supported due to the usage of Python libraries)
@gulsumgudukbay: you are searching for native GPU application or also tensorflow application?
1 reply
Hey everyone, this PR is a very good opportunity for a first-time contributor: galaxyproject/tools-iuc#4492
if you want to contribute this tool is easy to handle and needs an update
you need to add the missing parameters of the new version
I was trying to figure out why most of the tool tests of all versions of Pangolin 3.x fail on usegalaxy.org with:
AttributeError in line 112 of /usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/site-packages/pangolin/scripts/pangolearn.smk:
'DecisionTreeClassifier' object has no attribute 'n_features_'
File "/usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/site-packages/pangolin/scripts/pangolearn.smk", line 112, in __rule_pangolearn
File "/usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/site-packages/pangolin/pangolearn/pangolearn.py", line 170, in assign_lineage
File "/usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/site-packages/sklearn/tree/_classes.py", line 922, in predict_proba
File "/usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/site-packages/sklearn/tree/_classes.py", line 395, in _validate_X_predict
File "/usr/local/tools/_conda/envs/mulled-v1-72f7a2d4de377591b330113007411515001882bdf41ef6a774876b7e70240654/lib/python3.8/concurrent/futures/thread.py", line 57, in run
Exiting because a job execution failed. Look above for error messageBoth conda and singularity, and so wanted to see how usegalaxy.eu had fixed it - turns out, it's broken on EU too.
This seems to imply it's an install issue but I suspect reinstalling the deps isn't going to help in this case: cov-lineages/pangolin#427
fwiw I also get the following with pangolin 4.0.5+galaxy1 on EU (I don't have 4.x on .org):
Error: the input query fasta could not be parsed.
Double check your query fasta and that compressed stdin was not passed.
Please enter your fasta sequence file and refer to pangolin usage at: https://cov-lineages.org/pangolin.html for detailed instructions.
Error: the input query fasta could not be parsed.
Double check your query fasta and that compressed stdin was not passed.
Please enter your fasta sequence file and refer to pangolin usage at: https://cov-lineages.org/pangolin.html for detailed instructions.
rm: cannot remove '/data/jwd/main/046/882/46882007/tmp/out.3651279': No such file or directory
rm: cannot remove '/data/jwd/main/046/882/46882007/tmp/err.3651279': No such file or directory
I've been trying to explain to them why it's not a great idea to switch your scikit-learn version when building models for a given release, but seems they rae not really getting the point :(
That's for your 3.1.20 question.
Older models installed via a DM are working still.
Regarding the v4 series, there should be
4.0.5+galaxy2 in the toolshed already (not sure why it didn't get installed on .eu over the weekend), which addresses a few issues. Your error message really looks like something's wrong with your FASTA input though. If not, then what were the exact parameters you've been using?
Hi, I opened this PR to update rpbasicdesign tool : https://github.com/brsynth/tools-iuc/pull/new/rpbasicdesign_update_1_1_0 . Can anyone review/merge it please ? I also opened this PR (galaxyproject/tools-iuc#4458) for another tool (dnabot) and it was merged but I can't find it in the toolshed . Can you check what's wrong please? Thank you for your work.
Hope all is ok on your sides !
We are trying to add some conda recipes but tests fails...
If someone has an idea conda-forge/staged-recipes#18967
1 reply
> library('dissUtils') 2022-05-17T07:43:35.5948057Z Error in library("dissUtils") : there is no package called 'dissUtils'
(the recipes are related to
mmand, dissUtils and geometa R packages)
Moreover is there anyone there having created a conda recipe from an archived (outdated) R package?
just using https://github.com/bgruening/conda_r_skeleton_helper to generate conda recipe from R package, we have some "extra files" not mandatory normally for proposing a recipe BUT in fact this is mandatory for the github tests to b ok ;)
Another question ;)
concerning the fact we want a tool proposing as input data a composite format, shp.
we wrote this :
<inputs>
<param type="data" name="input" format="shp,xml,json,tabular,txt"https://toolshed.g2.bx.psu.edu/repository?repository_id=4127eedef3aaafec
but in fact I see that we can't access shp composite files from the tool (for example on eu: https://ecology.usegalaxy.eu/root?tool_id=toolshed.g2.bx.psu.edu/repos/ecology/gdal_ogr2ogr/gdal_ogr2ogr/3.0.0)
if you have any information
;)
hope all is ok on your side !
for example, with the ogr2ogr tool, I can't select the dataset 21
with type binary
so this is corerct that you can not select it