def test_GivenInconsistentClusterNumbers_RaisesError(self):
"""
`gapped_clusters` can occur when number of distinct data points < requested number of
clusters in kmeans, leading to production of empty clusters and then
messing up prg construction.
`large_clusters` should never occur, cluster assigment should be consecutive integers from 0.
"""
gapped_clusters = [0, 2, 2]
with self.assertRaises(ValueError):
extract_clusters(self.seqdict_ids, gapped_clusters)
large_clusters = [5, 6, 6]
with self.assertRaises(ValueError):
extract_clusters(self.seqdict_ids, large_clusters)
def test_GivenGroupedClusters_ExtractCorrectSequenceClusters(self):
cluster_assignment = [1, 1, 0]
actual = extract_clusters(self.seqdict_gapped_seqs, cluster_assignment)
expected = [["GGG"], ["A-T", "AT-", "TTT"]]
self.assertEqual(actual, expected)
def test_GivenGroupedClusters_ExtractCorrectIDClusters(self):
cluster_assignment = [0, 0, 1]
actual = extract_clusters(self.seqdict_ids, cluster_assignment)
expected = [["s1", "s2", "s3"], ["s4"]]
self.assertEqual(actual, expected)
def test_GivenTooFewClusterAssignments_Fails(self):
cluster_assignment = [0, 1]
for seqdict in [self.seqdict_ids, self.seqdict_gapped_seqs]:
with self.assertRaises(ValueError):
extract_clusters(seqdict, cluster_assignment)