Closed orbeckst closed 6 years ago
The hole-orderparameter.ipynb notebook refuses to run as originally.
ref = mda.Universe(PDB_HOLE) # reference structure
u = mda.Universe(MULTIPDB_HOLE) # trajectory
R = RMSD(u, reference=ref, select="protein", weights="mass")
R.run()
fails with (cryptic!)
MDAnalysis.analysis.rmsd: INFO RMS calculation for 260 atoms.
MDAnalysis.analysis.rmsd: ERROR Atoms: reference | mobile
MDAnalysis.analysis.rmsd: ERROR Inconsistent selections, masses differ by more than0.100000; mis-matching atomsare shown above.
I can add a script to install while I think. I can check during the week.
I mean install a script to install hole.
https://mybinder.org/v2/gh/MDAnalysis/binder-notebook/hole-notebooks
@orbeckst I added an install script. Maybe it works. The link above is the binder for the new branch to test.
@orbeckst I have no idea what the error message is btw. If hole is missing we should have a better error message.
The hole-basics
notebook works. The hole-orderparameter
notebook fails due to changes in the library. It would be beneficial to test the notebooks with releases.
Thanks for adding the installation script. It worked for me: I was able to run all of https://hub.mybinder.org/user/mdanalysis-binder-notebook-ymw4ep1z/notebooks/analysis/hole-basics.ipynb without issues.
I am not sure what error message you're referring to.
The second notebook https://hub.mybinder.org/user/mdanalysis-binder-notebook-ymw4ep1z/notebooks/analysis/hole-orderparameter.ipynb is failing but not because of hole.
The cryptic error you had in https://github.com/MDAnalysis/binder-notebook/pull/1#issuecomment-364784645
Yes – that's not hole but other stuff. Not sure what we changed to make it fail now. Need to investigate.
(And it fails locally for me, too.)
I can split the PR and we only put up the working notebook for right now. Should we do that?
Yeah that would be good. I think we also need to clean the install script. The artifacts that the hole installation leaves should be cleaned as they can be confusing.
On 15. Feb 2018, at 23:26, Oliver Beckstein notifications@github.com wrote:
I can split the PR and we only put up the working notebook for right now. Should we do that?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.
Yes & yes.
On Feb 15, 2018, at 3:32 PM, Max Linke notifications@github.com wrote:
Yeah that would be good. I think we also need to clean the install script. The artifacts that the hole installation leaves should be cleaned as they can be confusing.
PR https://github.com/MDAnalysis/mdanalysis/pull/1783 will fix the problem with the second notebook and it will work with MDAnalysis 0.17.1.
I decided to add both notebooks. The second one has a comment that it needs MDA 0.17.1 (but it's not very conspicuous). I hope we'll get PR MDAnalysis/mdanalysis#1783 merged soon and 0.17.1 released, then I don't have to worry about.
I can open an issue saying that the notebook is not full working once this PR is merged. At least in this way there's an official acknowledgment and we can easily close the issue once 0.17.1 is out.
I remove the hole tar file. I am not sure how to test this – can I run binder on a branch?
can I run binder on a branch?
See: https://github.com/MDAnalysis/binder-notebook/pull/1#issuecomment-366079295
D'oh. Thanks.
The downloaded hole tar file is gone. The new ~/hole2
directory might be a bit confusing but we might have to think a bit more how to organize the file system anyway. Perhaps just put all notebooks into a directory "notebooks" to make clear where to look?
I am inclined to merge it now, even though the hole-orderparameter notebook will only start working properly with MDA 0.17.1.
Any other opinions?
Started on the original gists.
However, for these notebooks to work we need a conda package for hole https://github.com/MDAnalysis/mdanalysis/issues/1429 so that it can be installed in Binder.