Figure 2.
Sample MSM code. MSMBuilder balances a powerful API with ease of use. A sample workflow is shown here using the Python API. Following the successful model of the broadly applicable scikit-learn package, each modeling step is represented by an estimator object that operates on the data. Here, the AlphaAngleFeaturizer transforms raw coordinates into a angles. The output of this transformation is fed into the tICA dimensionality reduction, the MiniBatchKMeans clustering algorithm, and finally the ContinuousTimeMSM model. MSMBuilder provides a litany of utility functions for dealing with large MD data sets for input and output. Although this example shows the Python API, MSMBuilder is fully functional from the command line with an intuitive one-to-one correspondence between Python estimator objects and command-line commands. To see this figure in color, go online.