def test_calculate_fromts(self):
matrix = DTW(self.xts, self.yts).calculate_cost()[0]
acc_cost = [[1., 1., 1., 2., 6., 7., 7.],
[2., 1., 1., 2., 6., 7., 7.],
[6., 2., 2., 1., 2., 2., 3.],
[15., 6., 6., 2., 1., 2., 6.],
[19., 7., 7., 2., 2., 1., 2.],
[19., 8., 8., 6., 11., 5., 2.]]
acc_cost = numpy.matrix(acc_cost)
self.assertEqual(matrix.all(), acc_cost.all())
def test(self):
CTGF = TimeSeries(self.CTGF)
smad7 = TimeSeries(self.smad7)
CTGF.interpolate(inplace=True)
smad7.interpolate(inplace=True)
CTGF.norm(inplace=True)
smad7.norm(inplace=True)
dtw = DTW(CTGF, smad7)
fig = dtw.cost_plot()
fig = dtw.plot()
plt.show()
self.assertTrue(isinstance(fig, Figure))
def dist_matrix(self):
dct = {}
matrix = numpy.zeros((len(self.tsg), len(self.tsg)))
matrix[:] = numpy.nan
df = pandas.DataFrame(matrix, columns=self.tsg.features, index=self.tsg.features)
comb = [i for i in combinations(self.tsg.features, 2)]
for i, j in comb:
cost = DTW(self.tsg[i], self.tsg[j]).cost
df.loc[i, j] = cost
df.loc[j, i] = cost
return df
def test2(self):
dir = r'/home/b3053674/Documents/Microarray/GSS2265/python'
tsg1 = TimeSeriesGroup(self.data.iloc[:10])
tsg1 = tsg1.norm()
tsg1.interpolate('linear', 30)
ts_l = tsg1.to_ts()
ts = ts_l[0]
ts2 = ts_l[2]
fname_plot = os.path.join(dir, 'CTGFVsCTGF.png')
fname_map = os.path.join(dir, 'CTGFVsCTGFMap.png')
fname_normed_plot = os.path.join(dir, 'CTGFVsCTGFnormed.png')
fname_normed_map = os.path.join(dir, 'CTGFVsCTGFnormedMap.png')
fname_normed_interp_plot = os.path.join(dir, 'CTGFVsCTGFnormedInterp.png')
fname_normed_interp_map = os.path.join(dir, 'CTGFVsCTGFnormedInterpMap.png')
d = DTW(ts, ts2)
fig = d.plot()
fig.savefig(fname_normed_interp_plot, dpi=300, bbox_inches='tight')
fig2 = d.cost_plot()
fig2.savefig(fname_normed_interp_map, dpi=300, bbox_inches='tight')
def compute_dtw(self, y):
return DTW(self.x, y)
def compute_dtw(self, vec):
x = self.clusters[vec[0]].centroid_by_eucl
y = self.clusters[vec[1]].centroid_by_eucl
dtw = DTW(x, y, dist=self.dtw_dist)
return vec[0], vec[1], dtw.cost
def dtw_wrapper(x, y, axis=1):
return DTW(x, y).cost
df = pandas.read_excel(os.path.join(dire, 'MicroarrayDEGAgeravedData.xlsx'), index_col=[0, 1])
df = df.transpose()
df = df['TGFb'] / df['Control']
# put all other information in db as well so that plots can be produced from reading it
perm = [i for i in itertools.combinations(list(df.index), 2)]
DOALL = True
if DOALL:
for x_gene, y_gene in perm:
dwt = DTW(df.loc[x_gene], df.loc[y_gene])
path = pickle.dumps(dwt.path)
db.execute("""INSERT INTO dwt_data ( x, y, cost, path)
VALUES ('{}', '{}', {}, '{}')""".format(
x_gene, y_gene, dwt.cost, path
)
)
db.commit()
# dtw1pair()
def test_from_df_list(self):
l = [DTW(self.CTGF, self.smad7),
DTW(self.smad7, self.CTGF)]
df = pandas.DataFrame(l)
self.assertTrue(isinstance(df.iloc[0, 0].cost, float))
def test_real_data(self):
dtw = DTW(self.CTGF, self.smad7)
self.assertTrue(isinstance(dtw, DTW))
def test_distance_cost_plot(self):
dtw = DTW(self.xts, self.yts)
fig = dtw.cost_plot()
from matplotlib.figure import Figure
self.assertTrue(isinstance(fig, Figure))
def test_find_best_path_fromts(self):
path, cost = DTW(self.xts, self.yts).find_best_path()
path_answer = [[0, 0], [0, 1], [1, 1], [1, 2], [2, 3], [3, 4], [4, 5], [5, 6]]
cost_ans = 2.0
self.assertListEqual(path_answer, path)
self.assertEqual(cost_ans, cost)
def test_same_example_from_DTW(self):
x = numpy.array([1, 1, 2, 3, 2, 0])
y = numpy.array([0, 1, 1, 2, 3, 2, 1])
dtw1 = DTW(x, y)
dtw2 = FastDTW(x, y, radius=1)
self.assertAlmostEqual(dtw1.cost, dtw2.cost)