def train(protein,clf,dim_red=None,encoding=False,single_frame=False):
'''
Args:
- protein: "bpti" or "alanine", name of the protein
- clf: classifier to use
- dim_red: int or None.
if not None, it's the reduced dimension for PCA
- encode: if False, classify on coordinates;
if True, classify on encoded hidden states
- single_frame: classify as single frames or sequences.
'''
# get features
if encoding==False:
# get flattened coordinates of each frame
traj = dr.load_traj(protein)
coords = traj.xyz
coords = np.reshape(coords,(len(coords),-1))
X = coords
# print coords
else:
# get encoded hidden states
X = encode()
# use PCA to reduce dimension
if dim_red != None:
pca = PCA(dim_red)
X = pca.fit(X).transform(X)
# print X[0]
# get labels
with open("/output/"+protein+"/labels"+suffix,"r") as lb:
# with open("/protein/data/"+protein+"-labels"+suffix) as lb:
Y = np.asarray(pickle.load(lb))
if encoding==True and single_frame==True:
# need to match the number of sequences
end = -seq_size*window_size+1
if end == 0:
end = Y.size
Y = Y[0:end:sliding*window_size] # compare with the first frame
# Y = Y[-end::sliding*window_size] # compare with the last frame
# print Y
# training and cross validation
data_scores = cross_val_score(clf,X,Y,cv=int(fold))
print("Accuracy with %s folds: %0.2f (+/- %0.2f)" %
(fold, data_scores.mean(), data_scores.std()))
centers = np.array([[55,48],[-77,138],[-77, -39],[60, -72]])*np.pi/180.0
clu = kmeans(n_clusters=4,init=centers)
# labels = clu.fit_predict(dihedrals)
labels = clu.fit_predict(np.transpose(dihedrals_shift))
labels =
print("centers:")
print(clu.cluster_centers_*180.0/np.pi)
return labels
if __name__ == "__main__":
task = int(sys.argv[1])
if task==1:
# do clustering and save the label file
traj = dr.load_traj(protein)
# labels = k_means(traj,num_states,dim_red=10)
labels = hierarchy(traj,num_states)
with open("/output/"+protein+"/labels"+suffix,"w") as lb:
pickle.dump(labels,lb)
elif task==2:
# read the label file and get the indices of one cluster
# with open("/output/"+protein+"/labels"+suffix,"r") as lb:
with open("/protein/data/"+protein+"-labels"+suffix) as lb:
labels = pickle.load(lb)
# for l in labels:
# print(l)
elif task==3:
# cluster the sequences