- Ported code to use MDTraj. This eliminates the need for MSMBuilder to
directly handle the loading and saving of trajectories, and gives us
much more cross-format support.
- New sphinx documentation
- Migrated from SVN to Github for version control. The source code can
now be found at https://github.com/SimTk/msmbuilder
- Renamed all the functions in MSMLib to be pep8 compliant.
Aliases have been added for the old names. Note: Trajectory.py is
still not pep8 compliant, but will be fixed by release 3.0.
- Split MSMLib into 2 files: MSMLib and msm_analysis. msm_analysis contains
all the code for analyzing MSMs (eigenvector calculation, sampling, etc).
MSMLib contains code for building models, while msm_analysis contains
code for working with models.
- Removed AssignMPI.py
- Added more unit tests and Travis automated unit testing.
- Updated most of MSMBuilder to be PEP8 compliant.
- Rewrote code for estimating reversible maximum likelihood count matrix.
The new code should be faster and easier to read.
- New code implementing min-cut/max-flow cut-based reaction coordinate
learning from MSMs (cfep.py). Beta version.
- New code for the computation of MSM hub scores.
- Switched to YAML for storing project information.
- Added script RebuildProject.py to construct YAML ProjectInfo file.
- Deleted the Serializer class.
- New msmbuilder.io module for storing output.
- Added support for the BACE coarse graining algorithm.
- Better error checking for assignments arrays.
- Updates to tutorials.
- Rewrote lumping code for improved readability.
- Updates to tutorials.
- Moved Ward clustering and SCRE to AdvancedMethods tutorial.
- The libraries are distance metric agnostic, you can use your own
new distance metric without having to change the clustering/assignment code.
- Dihedral, Contact (residues and/or atoms), hybrid distance metric code
- New clustering algorithms (CLARANS / subsampled CLARANS)
- Flux-based PCCA+.
- Support for Hierarchical Clustering (e.g. Ward).
- SCRE rate matrix estimation.
- Improvements in PCCA, PCCA+ lead to better macrostate definition.
- Bug fix: In ConvertDataToHDF, the Stride option was not being used
for non-FAH style datasets.
- Fixed unicode strings were causing scipy issues on certain platforms.
- Fixed issue with AtomIndices in UpdateProjectToHDF
- Using dtype=’int’ for Assignments, consistent throughout MSMBuilder.
- CalculateImpliedTimescales now allows users to exit through control+C
- Fixed some tabbing issues in MSMLib.py
- Fixed a bug in ConvertProjectToHDF that prevented MSMBuilder from seeing
multiple XTC files in a single directory.
- Fixed carriage return issues in UpdateProjectToHDF
- Fixed minor bugs in Project class.
- If Project cannot find its PDB, it also will look in the current directory.
This helps when the absolute path of a Project changes.
- Unit tests.
- Removed unnecessary files, reducing the package size to 10MB.