def classifyWorm(labels, wid, d, s, nbins = 256, verbose = True):
XYwdata = XYdata[wid].copy();
w = wd.WormData(XYwdata[:,0:2], stage = XYwdata[:,-1], valid = XYwdata[:,0] != 1, label = ('x', 'y'), wid = wid);
w.replaceInvalid();
ds = w.calculateDistances(n = delays[d]+1, stage = s);
rs = w.calculateRotations(n = delays[d]+1, stage = s);
ddata = np.vstack([np.log(ds[:,-1]), (rs[:,-1])]).T
#gmmdata = np.vstack([dists[:,j], (rots[:,j])]).T
#ddata.shape
nanids = np.logical_or(np.any(np.isnan(ddata), axis=1), np.any(np.isinf(ddata), axis = 1));
ddata = ddata[~nanids,:];
#ddata.shape
pred2 =-np.ones(rs.shape[0])
pred2nn = pred2[~nanids];
pred2nn.shape
for i in range(2):
ddata[:,i] = ddata[:,i] - ddata[:,i].min();
ddata[:,i] = (ddata[:,i] / ddata[:,i].max()) * (nbins-1);
ddata = np.asarray(ddata, dtype = int);
for i in range(2):
ddata[ddata[:,i] > (nbins-1), i] = nbins-1;
for i in xrange(ddata.shape[0]):
pred2nn[i] = labels[ddata[i,0], ddata[i,1]];
pred2[~nanids] = pred2nn;
#pred2nn.max();
if verbose:
plt.figure(506); plt.clf();
w.plotTrace(ids = shiftData(pred2, delays[d]/2, nan = -1), stage = s)
if verbose > 2:
rds = w.calculateRotations(n = delays[d] + 1, stage = s);
plt.figure(510); plt.clf();
w.plotDataColor(data = shiftData(rds[:, -1], delays[d]/2, nan = -1), c = pred2, lw = 0, s = 20, stage = s, cmap = cm.rainbow)
dts = w.calculateDistances(n = delays[d] + 1, stage = s);
plt.figure(510); plt.clf();
w.plotDataColor(data = dts[:, -1], c = pred2, lw = 0, s = 20, stage = s, cmap = cm.rainbow)
plt.figure(507); plt.clf();
w.plotTrajectory(stage = s, colordata = shiftData(rds[:,-1] /(delays[d]+1) /np.pi, delays[d]/2, nan = -.1))
dist2 = w.calculateLengths(n=200);
plt.figure(511); plt.clf();
w.plotDataColor(data = dist2[:, -1], c = pred2, lw = 0, s = 20)
return pred2;
def plotData(data, iters = [0,1,10,20], name = None):
#plt.figure(10); plt.clf();
#plt.plot(np.log(range(1,nn+1)), np.log(rots[3200:3300,:].T))
#fig = plt.figure(10); plt.clf();
n = len(iters);
n2 = int(np.ceil(n/2.));
fig, axs = plt.subplots(2, n2, sharex=True, sharey=True);
for k,i in enumerate(iters):
#plt.subplot(2,2,k+1);
w.plotTrajectory(colordata = shiftData(data, i/2, reverse = True)[:,i], ax = axs.flat[k]);
axs.flat[k].set_title(str(i));
if name is not None:
fig = plt.gcf();
fig.canvas.set_window_title(name);
fig = plt.figure(); plt.clf();
for k,i in enumerate(iters):
plt.subplot(2,n2,k+1);
plt.hist(data[~np.isnan(data[:,i]),i], bins = 50)
plt.title(str(i));
if name is not None:
fig = plt.gcf();
fig.canvas.set_window_title(name);
def classifyWorm(labels, wid, d, s, nbins=256, verbose=True):
XYwdata = XYdata[wid].copy()
w = wd.WormData(XYwdata[:, 0:2],
stage=XYwdata[:, -1],
valid=XYwdata[:, 0] != 1,
label=('x', 'y'),
wid=wid)
w.replaceInvalid()
ds = w.calculateDistances(n=delays[d] + 1, stage=s)
rs = w.calculateRotations(n=delays[d] + 1, stage=s)
ddata = np.vstack([np.log(ds[:, -1]), (rs[:, -1])]).T
#gmmdata = np.vstack([dists[:,j], (rots[:,j])]).T
#ddata.shape
nanids = np.logical_or(np.any(np.isnan(ddata), axis=1),
np.any(np.isinf(ddata), axis=1))
ddata = ddata[~nanids, :]
#ddata.shape
pred2 = -np.ones(rs.shape[0])
pred2nn = pred2[~nanids]
pred2nn.shape
for i in range(2):
ddata[:, i] = ddata[:, i] - ddata[:, i].min()
ddata[:, i] = (ddata[:, i] / ddata[:, i].max()) * (nbins - 1)
ddata = np.asarray(ddata, dtype=int)
for i in range(2):
ddata[ddata[:, i] > (nbins - 1), i] = nbins - 1
for i in xrange(ddata.shape[0]):
pred2nn[i] = labels[ddata[i, 0], ddata[i, 1]]
pred2[~nanids] = pred2nn
#pred2nn.max();
if verbose:
plt.figure(506)
plt.clf()
w.plotTrace(ids=shiftData(pred2, delays[d] / 2, nan=-1), stage=s)
if verbose > 2:
rds = w.calculateRotations(n=delays[d] + 1, stage=s)
plt.figure(510)
plt.clf()
w.plotDataColor(data=shiftData(rds[:, -1], delays[d] / 2, nan=-1),
c=pred2,
lw=0,
s=20,
stage=s,
cmap=cm.rainbow)
dts = w.calculateDistances(n=delays[d] + 1, stage=s)
plt.figure(510)
plt.clf()
w.plotDataColor(data=dts[:, -1],
c=pred2,
lw=0,
s=20,
stage=s,
cmap=cm.rainbow)
plt.figure(507)
plt.clf()
w.plotTrajectory(stage=s,
colordata=shiftData(rds[:, -1] / (delays[d] + 1) /
np.pi,
delays[d] / 2,
nan=-.1))
dist2 = w.calculateLengths(n=200)
plt.figure(511)
plt.clf()
w.plotDataColor(data=dist2[:, -1], c=pred2, lw=0, s=20)
return pred2
w = wd.WormData(xyd, stage = np.ones(nn), valid = np.ones(nn, dtype = bool), label = ('x', 'y'), wid = wid);
w.replaceInvalid();
w.plotTrajectory()
rads = [0.1, 0.5, 1., 2]
npath = 4;
pe = w.calculatePathEntropy(radius = rads, n = npath)
print pe
plt.clf();
nr = pe.shape[1];
for i in range(nr):
plt.subplot(nr,1,i+1);
w.plotTrace(shiftData(pe[:,i], s = npath/2));
plt.title('path entropy radius = %f' % rads[i])
## test on real data
basedir = '/home/ckirst/Science/Projects/CElegansBehaviour/';
filename = os.path.join(basedir, 'Experiment/individuals_N2_X_Y.mat')
data = io.loadmat(filename);
XYdata = data['individual_X_Y'][0];
print XYdata.shape
print XYdata[0].shape
angs = w.calculateAngles(n = nn); #plt.figure(10); plt.clf(); #plt.plot(range(1,nn+1), angs[3200:3300,:].T) plotData(angs, iters = nns, name = 'Angles') ############## #diffusions nn = 50; diffs = w.calculateDiffusionParameter(n=nn, parallel = True) plt.figure(20); plt.clf(); diffsshift = shiftData(diffs, nn/2, reverse = True); w.plotTrajectory(colordata = diffsshift[:,0], size = 100) plt.figure(21); plt.clf(); w.plotTrajectory(colordata = diffsshift[:,0], size = diffsshift[:,-1]* 1000) plt.figure(22); plt.clf(); w.plotTrajectory(colordata = diffsshift[:,1], size = 60) #w.plotTrajectory(colordata = -diffsshift[:,-1], size = diffsshift[:,0]* 1000) #histogram nbins = 50; plt.figure(13); plt.clf(); plt.subplot(2,2,1); plt.hist(diffs[~np.isnan(diffs[:,0]),0], bins = nbins) plt.subplot(2,2,2); plt.hist(diffs[~np.isnan(diffs[:,0]),-1], bins = nbins)
ddata[:,i] = ddata[:,i] - ddata[:,i].min(); ddata[:,i] = (ddata[:,i] / ddata[:,i].max()) * (nbins-1); ddata = np.asarray(ddata, dtype = int); for i in range(2): ddata[ddata[:,i] > (nbins-1), i] = nbins-1; for i in xrange(ddata.shape[0]): pred2nn[i] = labelsws[ddata[i,0], ddata[i,1]]; pred2[~nanids] = pred2nn; pred2nn.max(); plt.figure(506); plt.clf(); w.plotTrace(ids = shiftData(pred2, delays[d]/2, nan = -1), stage = s) rds = w.calculateRotations(n = delays[d] + 1, stage = s); plt.figure(510); plt.clf(); w.plotDataColor(data = shiftData(rds[:, -1], delays[d]/2, nan = -1), c = pred2, lw = 0, s = 20, stage = s, cmap = cm.rainbow) dts = w.calculateDistances(n = delays[d] + 1, stage = s); plt.figure(510); plt.clf(); w.plotDataColor(data = dts[:, -1], c = pred2, lw = 0, s = 20, stage = s, cmap = cm.rainbow) plt.figure(507); plt.clf(); w.plotTrajectory(stage = s, colordata = shiftData(rds[:,-1] /(delays[d]+1) /np.pi, delays[d]/2, nan = -.1))
wid=wid)
w.replaceInvalid()
w.plotTrajectory()
rads = [0.1, 0.5, 1., 2]
npath = 4
pe = w.calculatePathEntropy(radius=rads, n=npath)
print pe
plt.clf()
nr = pe.shape[1]
for i in range(nr):
plt.subplot(nr, 1, i + 1)
w.plotTrace(shiftData(pe[:, i], s=npath / 2))
plt.title('path entropy radius = %f' % rads[i])
## test on real data
basedir = '/home/ckirst/Science/Projects/CElegansBehaviour/'
filename = os.path.join(basedir, 'Experiment/individuals_N2_X_Y.mat')
data = io.loadmat(filename)
XYdata = data['individual_X_Y'][0]
print XYdata.shape
print XYdata[0].shape
nworms = XYdata.shape[0]