Closed rmcgibbo closed 10 years ago
I'm going to tag a release of the current master as 2.7.0 before this merge happens.
Good On Jan 21, 2014 9:40 PM, "Robert McGibbon" notifications@github.com wrote:
I'm going to tag a release of the current master as 2.7.0 before this merge happens.
— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/pull/292#issuecomment-32988586 .
Let's get some comment on this from a quorum of @tjlane, @schwancr, @mpharrigan, etc before merging this. There's no rush.
A quick scan looks good to me, but I admit I didn't delve into many details. Anything in particular you want reviewed @rmcgibbo since I'm not really working with MSMb right now?
+1 to getting mdtraj in
Here's my suggestion of what to look for (without having to invest 10 hours):
So Danny and I will be doing some MSM-building, so we'll be working with this pretty extensively once it's merged. I think we should be able to find and patch any bugs that are still hidden.
OK so I've merged all my "minor" fixes and enhancements into the msmb2.8 branch. If people want to see what th changes are, they can look at the various merged Pull Requests.
It seems like testAssign just hangs sometimes? https://travis-ci.org/SimTk/msmbuilder/builds/17891035#L1451
There are a couple of sporatic problems with the travis build system, but this PR is in a working state now. We should merge. I will do it tomorrow afternoon.
mdtraj uses simtk.units, is there some way to install just the units module? Or do I need to install all of openmm?
I think Robert has a way to install just the units part.
git+git://github.com/rmcgibbo/simtk.unit
cool, thanks
is it possible to install that with pip? It would be nice to have that dependency installed automatically
I'm using this branch now, so I can be the guinea pig if there are bugs
I'm not sure exactly why simtk.units is required by msmbuilder2.8. In general, most of mdtraj should be okay without it.
Or maybe I should just bundle a copy of simtk.units inside mdtraj.
-Robert
On Jan 30, 2014, at 3:43 PM, Christian Schwantes notifications@github.com wrote:
I'm using this branch now, so I can be the guinea pig if there are bugs
— Reply to this email directly or view it on GitHub.
here's the traceback:
In [2]: md.load('Trajectories/trj0.h5')
############################################################################
The code at unit.py:147 requires the simtk.unit module,
which is a python package for unit conversion.
simtk.unit is installed with OpenMM, which is available at http://openmm.org
It's also installable as a separate standalone package from
https://github.com/rmcgibbo/simtk.unit, and can be installed with the python
"pip" package mangers using:
pip install git+git://github.com/rmcgibbo/simtk.unit
############################################################################
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
/home/schwancr/vspm24_vol/vspm24_HD2/protG/fah_nug2/p7812/<ipython-input-2-f2dfa16a7d2d> in <module>()
----> 1 md.load('Trajectories/trj0.h5')
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/trajectory.pyc in load(filename_or_filenames, discard_overlapping_frames, **kwargs)
282 kwargs.pop('top', None)
283
--> 284 return loader(filename, **kwargs)
285
286
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/hdf5.pyc in load_hdf5(filename, stride, atom_indices, frame)
149 data = f.read(n_frames=1, atom_indices=atom_indices)
150 else:
--> 151 data = f.read(stride=stride, atom_indices=atom_indices)
152
153 topology = f.topology
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/hdf5.pyc in wrapper(*args, **kwargs)
84 # args[0] is self on the method
85 if args[0].mode in m:
---> 86 return f(*args, **kwargs)
87 raise ValueError('This operation is only available when a file '
88 'is open in mode="%s".' % args[0].mode)
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/hdf5.pyc in read(self, n_frames, stride, atom_indices)
604 frames = Frames(
605 coordinates = get_field('coordinates', (frame_slice, atom_slice, slice(None)),
--> 606 out_units='nanometers', can_be_none=False),
607 time = get_field('time', frame_slice, out_units='picoseconds'),
608 cell_lengths = get_field('cell_lengths', (frame_slice, slice(None)), out_units='nanometers'),
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/hdf5.pyc in get_field(name, slice, out_units, can_be_none)
595 if not isinstance(in_units, str):
596 in_units = in_units.decode()
--> 597 data = in_units_of(data, out_units, in_units)
598 return data
599 except self.tables.NoSuchNodeError:
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/utils/unit.pyc in in_units_of(quantity, units_out, units_in)
145 1000000.0
146 """
--> 147 units = import_('simtk.unit')
148
149 if quantity is None:
/home/schwancr/Installed/epd/lib/python2.7/site-packages/mdtraj-0.6.0-py2.7-linux-x86_64.egg/mdtraj/utils/delay_import.pyc in import_(module)
191 print(m, file=sys.stderr)
192 print(bar, file=sys.stderr)
--> 193 raise e
ImportError: No module named simtk.unit
I'm not all that familiar with mdtraj to know if it needs to do that conversion step
You can now install simtk.unit with
pip install simtk.unit
Sweet, let me know if there's anything else we need to do before we merge this.
Okay. I'm starting the merge now.
Nice
Big merge.
This PR introduces the mdtraj dependency, and removes lots of cruft.