Closed mpharrigan closed 10 years ago
Thanks for taking care of this. Looks like a good start. I guess we still need to finalize "how" drastic the changes in metrics will be for 2.8, given that Robert and Christian have giant plans for MSMB3.
since we're deleting the entire msmbuilder.geometry module, could we add some sort of link or warning when someone tries to import it, and tell the user to use mdtraj.geometry?
I guess we still need to finalize "how" drastic the changes in metrics will be for 2.8, given that Robert and Christian have giant plans for MSMB3.
I think we should just get things working, but not worry about making it pretty since we will be switching to features rather than metrics.
Agreed.
I merged @schwancr 's changes into my PR.
I've started testing the changes. I've tested:
For Dihedral clustering to work, you need my PR for mdtraj as well
since @rmcgibbo is incapacitated at the moment, why don't we divvy up the remaining files. This shouldn't be too complicated. @mpharrigan can you make a list of files that you haven't done yet and then we can split the work?
So is the plan for 2.8 to still essentially require the use of HDF5 trajectories?
I think I've made a first pass at all the relevant files in src/
. There are still some Scripts (e.g. CalculateProjectRMSD) that I haven't done
At this point, I think we should start testing all the functionality to see where things break. Just fixing Cluster.py turned up a lot of issues I thought I had taken care of
I started working on this, and cleaning up the scripts. I also added a PR for a new md.iterload
on mdtraj which will make a couple things easier: https://github.com/rmcgibbo/mdtraj/pull/243.
Changing metrics/ and geometry/ to use mdtraj
Sorry again about eclipse fixing all the whitespace. Use diff -w