def calc_rmsd(struct1, struct2): crds1 = get_coords(struct1) crds2 = get_coords(struct2) assert(crds1.shape[1] == 3) assert(crds1.shape == crds2.shape) rmsd = rmsd_qcp(crds1, crds2) print "RMSD: %.4f" % rmsd
print 'Frames loaded successfully'
# concatenate all of the data together into a
# single array array
xyz = np.concatenate(xyz)
print 'Number of frames of data loaded %s' % len(xyz)
# compute the full pairwise distance matrix
# between all of the observed conformations
# using the rmsd_qcp algorithm
print 'Computing the pairwise distance matrix'
distances = np.empty(len(xyz)*(len(xyz)-1)/2)
k = 0
for i in xrange(len(xyz)):
for j in xrange(i+1, len(xyz)):
distances[k] = rmsd.rmsd_qcp(xyz[j], xyz[j])
k += 1
print 'RMSD computation %s/%s' % (k, len(xyz)*(len(xyz)-1)/2)
print 'mean pairwise distance ', np.mean(distances)
print 'stddev pairwise distance', np.std(distances)
print 'Running hierarchical clustering (UPGMA)...'
# run hierarchical clustering on the distance matrix
Z = linkage(distances, method='average')
# get flat clusters from the linkage matrix corresponding
# to states
print 'Flattening the clusters...'
assignments = fcluster(Z, t=5000, criterion='maxclust')
def calculate_rmsd(xyz1, xyz2, atoms=None):
return rmsd.rmsd_qcp(xyz1, xyz2)
