import scipy.optimize import mixtape.featurizer, mixtape.tica, mixtape.cluster, mixtape.markovstatemodel, mixtape.ghmm, mixtape.ntica import numpy as np import mdtraj as md from parameters import load_trajectories, build_full_featurizer import sklearn.pipeline, sklearn.externals.joblib import mixtape.utils stride = 1 trj0, trajectories, filenames = load_trajectories(stride=stride) trj0 = md.load("./system.subset.pdb") trj = trajectories[0] rmsd0 = md.rmsd(trj, trj0, 0) rmsd1 = md.rmsd(trj, trj, 0) mean(rmsd0[0:-1] * rmsd1[1:]) mean(rmsd1[0:-1] * rmsd0[1:]) X = [np.array([rmsd0, rmsd1]).T] ntica = mixtape.ntica.NtICA() #ntica._lambda = lam #ntica._lambda = np.array([-0.1, 0.1]) ntica.fit(X) ntica.offset_correlation_ X0, X1 = ntica.X0, ntica.X1 lam = np.zeros(2) Q0 = np.exp(-X0.dot(lam))
import mixtape.featurizer, mixtape.tica, mixtape.cluster, mixtape.markovstatemodel, mixtape.datasets, mixtape.subset_featurizer, mixtape.feature_selection import numpy as np import sklearn.pipeline, sklearn.externals.joblib import mixtape.utils from parameters import load_trajectories n_iter = 5000 n_choose = 10 lag_time = 1 trj0, trajectories, filenames = load_trajectories() train = trajectories featurizer = mixtape.subset_featurizer.guess_featurizers(trajectories[0], n_choose) tica_optimizer = mixtape.feature_selection.TICAOptimizer(featurizer, lag_time=lag_time) tica_optimizer.optimize(n_iter, train) #sklearn.externals.joblib.dump(tica_optimizer.featurizer, "./featurizer-%d.job" % n_choose, compress=True)