def reshape(readlen_dist, edges, heights):
lenDistShaper = functools.partial(_cont_dist_shaper, dist_shaper(
[(heights, edges)], nbins=40, trim_excess=False))
readlen_dist = lenDistShaper(readlen_dist)
nbins = readlen_dist.numBins
heights = readlen_dist.bins
bin_width = readlen_dist.binWidth
edges = [float(bn) * bin_width for bn in xrange(nbins)]
return edges, heights, bin_width
def test_abstract_dist_shaper_float_bwidth(self):
bins1 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0]
labels1 = [i * 0.2 for i in range(len(bins1))]
bins2 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
labels2 = [1.0 + i * 0.2 for i in range(len(bins2))]
dist_list = [(bins1, labels1), (bins2, labels2)]
merged = [0, 2, 3, 4, 3, 2, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
for nbins in range(1, 100):
shaper = dist_shaper(dist_list, nbins=nbins)
for dist in dist_list:
bins, labels = shaper(dist)
self.assertEqual(len(bins), len(labels))
self.assertEqual(len(bins), nbins)
self.assertEqual(sum(bins), sum(dist[0]))
"""
def test_abstract_dist_shaper_float_bwidth(self):
bins1 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0]
labels1 = [i * 0.2 for i in range(len(bins1))]
bins2 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
labels2 = [1.0 + i * 0.2 for i in range(len(bins2))]
dist_list = [(bins1, labels1), (bins2, labels2)]
merged = [0, 2, 3, 4, 3, 2, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
for nbins in range(1, 100):
shaper = dist_shaper(dist_list, nbins=nbins)
for dist in dist_list:
bins, labels = shaper(dist)
self.assertEqual(len(bins), len(labels))
self.assertEqual(len(bins), nbins)
self.assertEqual(sum(bins), sum(dist[0]))
"""
def test_abstract_dist_shaper(self):
bins1 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0]
labels1 = [i * 5 for i in range(len(bins1))]
bins2 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
labels2 = [25 + i * 5 for i in range(len(bins2))]
dist_list = [(bins1, labels1), (bins2, labels2)]
merged = [0, 2, 3, 4, 3, 2, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
for nbins in range(1, 100):
shaper = dist_shaper(dist_list, nbins=nbins)
for dist in dist_list:
# print 'pre'
# print dist[0]
bins, labels = shaper(dist)
# print 'post'
# print bins
# print labels
self.assertEqual(len(bins), len(labels))
self.assertEqual(len(bins), nbins)
self.assertEqual(sum(bins), sum(dist[0]))
# print ""
"""
def test_abstract_dist_shaper(self):
bins1 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0]
labels1 = [i * 5 for i in range(len(bins1))]
bins2 = [0, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
labels2 = [25 + i * 5 for i in range(len(bins2))]
dist_list = [(bins1, labels1), (bins2, labels2)]
merged = [0, 2, 3, 4, 3, 2, 2, 3, 4, 3, 2, 0, 0, 0, 0, 0, 0, 0]
for nbins in range(1, 100):
shaper = dist_shaper(dist_list, nbins=nbins)
for dist in dist_list:
# print 'pre'
# print dist[0]
bins, labels = shaper(dist)
# print 'post'
# print bins
# print labels
self.assertEqual(len(bins), len(labels))
self.assertEqual(len(bins), nbins)
self.assertEqual(sum(bins), sum(dist[0]))
# print ""
"""
def test_dist_shaper_leading_and_trailing(self):
# basic setup:
bins = [0, 0, 0, 0, 1, 2, 3, 4, 0, 0, 0, 0]
labels = [i * 5 for i in range(len(bins))]
self.assertEqual(len(bins), len(labels))
self.assertEqual(labels[-1], 55)
# no-op re-binning
shaper = dist_shaper([(bins, labels)], nbins=len(bins))
ob, ol = shaper((bins, labels))
self.assertEqual(bins, ob)
self.assertEqual(labels, ol)
# nbins = npopulated_columns - 1
eb = [3, 7, 0]
el = [20, 30, 40]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns - 1, excess trimmed
eb = [3, 7]
el = [20, 30]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns
eb = [1, 2, 3, 4]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins)
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns + 1
eb = [1, 2, 3, 4, 0]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins) + 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns + 1, excess trimmed
eb = [1, 2, 3, 4]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins) + 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# larger setup:
bins = [0, 0, 0, 0, 1, 2, 3, 4, 5, 0, 0, 0]
labels = [i * 5 for i in range(len(bins))]
self.assertEqual(len(bins), len(labels))
self.assertEqual(labels[-1], 55)
# nbins = npopulated_columns - 1
eb = [3, 7, 5, 0]
el = [20, 30, 40, 50]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns - 1, excess trimmed
eb = [3, 7, 5]
el = [20, 30, 40]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
def test_dist_shaper_leading_and_trailing(self):
# basic setup:
bins = [0, 0, 0, 0, 1, 2, 3, 4, 0, 0, 0, 0]
labels = [i * 5 for i in range(len(bins))]
self.assertEqual(len(bins), len(labels))
self.assertEqual(labels[-1], 55)
# no-op re-binning
shaper = dist_shaper([(bins, labels)], nbins=len(bins))
ob, ol = shaper((bins, labels))
self.assertEqual(bins, ob)
self.assertEqual(labels, ol)
# nbins = npopulated_columns - 1
eb = [3, 7, 0]
el = [20, 30, 40]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns - 1, excess trimmed
eb = [3, 7]
el = [20, 30]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns
eb = [1, 2, 3, 4]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins)
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns + 1
eb = [1, 2, 3, 4, 0]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins) + 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns + 1, excess trimmed
eb = [1, 2, 3, 4]
el = [i * 5 for i in range(4, 4 + len(eb))]
nbins = np.count_nonzero(bins) + 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# larger setup:
bins = [0, 0, 0, 0, 1, 2, 3, 4, 5, 0, 0, 0]
labels = [i * 5 for i in range(len(bins))]
self.assertEqual(len(bins), len(labels))
self.assertEqual(labels[-1], 55)
# nbins = npopulated_columns - 1
eb = [3, 7, 5, 0]
el = [20, 30, 40, 50]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
# nbins = npopulated_columns - 1, excess trimmed
eb = [3, 7, 5]
el = [20, 30, 40]
nbins = np.count_nonzero(bins) - 1
shaper = dist_shaper([(bins, labels)], nbins=nbins, trim_excess=True)
ob, ol = shaper((bins, labels))
self.assertEqual(eb, ob)
self.assertEqual(el, ol)
