def test_simple(n_children):
factory = GraphFactory(5, 100, 3)
root = SBT(factory, d=n_children)
leaf1 = Leaf("a", factory())
leaf1.data.count('AAAAA')
leaf1.data.count('AAAAT')
leaf1.data.count('AAAAC')
leaf2 = Leaf("b", factory())
leaf2.data.count('AAAAA')
leaf2.data.count('AAAAT')
leaf2.data.count('AAAAG')
leaf3 = Leaf("c", factory())
leaf3.data.count('AAAAA')
leaf3.data.count('AAAAT')
leaf3.data.count('CAAAA')
leaf4 = Leaf("d", factory())
leaf4.data.count('AAAAA')
leaf4.data.count('CAAAA')
leaf4.data.count('GAAAA')
leaf5 = Leaf("e", factory())
leaf5.data.count('AAAAA')
leaf5.data.count('AAAAT')
leaf5.data.count('GAAAA')
root.add_node(leaf1)
root.add_node(leaf2)
root.add_node(leaf3)
root.add_node(leaf4)
root.add_node(leaf5)
def search_kmer(obj, seq):
return obj.data.get(seq)
leaves = [leaf1, leaf2, leaf3, leaf4, leaf5]
kmers = ["AAAAA", "AAAAT", "AAAAG", "CAAAA", "GAAAA"]
def search_kmer_in_list(kmer):
x = []
for l in leaves:
if l.data.get(kmer):
x.append(l)
return set(x)
for kmer in kmers:
assert set(root.find(search_kmer, kmer)) == search_kmer_in_list(kmer)
print('-----')
print([x.metadata for x in root.find(search_kmer, "AAAAA")])
print([x.metadata for x in root.find(search_kmer, "AAAAT")])
print([x.metadata for x in root.find(search_kmer, "AAAAG")])
print([x.metadata for x in root.find(search_kmer, "CAAAA")])
print([x.metadata for x in root.find(search_kmer, "GAAAA")])
def test_simple_index(n_children):
factory = GraphFactory(5, 100, 3)
root = SBT(factory, d=n_children)
leaf1 = Leaf("a", factory())
leaf1.data.count("AAAAA")
leaf1.data.count("AAAAT")
leaf1.data.count("AAAAC")
leaf2 = Leaf("b", factory())
leaf2.data.count("AAAAA")
leaf2.data.count("AAAAT")
leaf2.data.count("AAAAG")
leaf3 = Leaf("c", factory())
leaf3.data.count("AAAAA")
leaf3.data.count("AAAAT")
leaf3.data.count("CAAAA")
leaf4 = Leaf("d", factory())
leaf4.data.count("AAAAA")
leaf4.data.count("CAAAA")
leaf4.data.count("GAAAA")
leaf5 = Leaf("e", factory())
leaf5.data.count("AAAAA")
leaf5.data.count("AAAAT")
leaf5.data.count("GAAAA")
root.add_node(leaf1)
root.add_node(leaf2)
root.add_node(leaf3)
root.add_node(leaf4)
root.add_node(leaf5)
def search_kmer(obj, seq):
return obj.data.get(seq)
kmers = ["AAAAA", "AAAAT", "AAAAG", "CAAAA", "GAAAA"]
linear = LinearIndex()
linear.insert(leaf1)
linear.insert(leaf2)
linear.insert(leaf3)
linear.insert(leaf4)
linear.insert(leaf5)
for kmer in kmers:
assert set(root.find(search_kmer,
kmer)) == set(linear.find(search_kmer, kmer))
print("-----")
print([x.metadata for x in root.find(search_kmer, "AAAAA")])
print([x.metadata for x in root.find(search_kmer, "AAAAT")])
print([x.metadata for x in root.find(search_kmer, "AAAAG")])
print([x.metadata for x in root.find(search_kmer, "CAAAA")])
print([x.metadata for x in root.find(search_kmer, "GAAAA")])
def test_longer_search(n_children):
ksize = 5
factory = GraphFactory(ksize, 100, 3)
root = SBT(factory, d=n_children)
leaf1 = Leaf("a", factory())
leaf1.data.count('AAAAA')
leaf1.data.count('AAAAT')
leaf1.data.count('AAAAC')
leaf2 = Leaf("b", factory())
leaf2.data.count('AAAAA')
leaf2.data.count('AAAAT')
leaf2.data.count('AAAAG')
leaf3 = Leaf("c", factory())
leaf3.data.count('AAAAA')
leaf3.data.count('AAAAT')
leaf3.data.count('CAAAA')
leaf4 = Leaf("d", factory())
leaf4.data.count('AAAAA')
leaf4.data.count('CAAAA')
leaf4.data.count('GAAAA')
leaf5 = Leaf("e", factory())
leaf5.data.count('AAAAA')
leaf5.data.count('AAAAT')
leaf5.data.count('GAAAA')
root.add_node(leaf1)
root.add_node(leaf2)
root.add_node(leaf3)
root.add_node(leaf4)
root.add_node(leaf5)
def kmers(k, seq):
for start in range(len(seq) - k + 1):
yield seq[start:start + k]
def search_transcript(node, seq, threshold):
presence = [node.data.get(kmer) for kmer in kmers(ksize, seq)]
if sum(presence) >= int(threshold * (len(seq) - ksize + 1)):
return 1
return 0
try1 = [x.metadata for x in root.find(search_transcript, "AAAAT", 1.0)]
assert set(try1) == set(['a', 'b', 'c', 'e']), try1 # no 'd'
try2 = [x.metadata for x in root.find(search_transcript, "GAAAAAT", 0.6)]
assert set(try2) == set(['a', 'b', 'c', 'd', 'e'])
try3 = [x.metadata for x in root.find(search_transcript, "GAAAA", 1.0)]
assert set(try3) == set(['d', 'e']), try3
def test_simple(n_children):
factory = GraphFactory(5, 100, 3)
root = SBT(factory, d=n_children)
leaf1 = Leaf("a", factory())
leaf1.data.count('AAAAA')
leaf1.data.count('AAAAT')
leaf1.data.count('AAAAC')
leaf2 = Leaf("b", factory())
leaf2.data.count('AAAAA')
leaf2.data.count('AAAAT')
leaf2.data.count('AAAAG')
leaf3 = Leaf("c", factory())
leaf3.data.count('AAAAA')
leaf3.data.count('AAAAT')
leaf3.data.count('CAAAA')
leaf4 = Leaf("d", factory())
leaf4.data.count('AAAAA')
leaf4.data.count('CAAAA')
leaf4.data.count('GAAAA')
leaf5 = Leaf("e", factory())
leaf5.data.count('AAAAA')
leaf5.data.count('AAAAT')
leaf5.data.count('GAAAA')
root.add_node(leaf1)
root.add_node(leaf2)
root.add_node(leaf3)
root.add_node(leaf4)
root.add_node(leaf5)
def search_kmer(obj, seq):
return obj.data.get(seq)
leaves = [leaf1, leaf2, leaf3, leaf4, leaf5]
kmers = ["AAAAA", "AAAAT", "AAAAG", "CAAAA", "GAAAA"]
def search_kmer_in_list(kmer):
x = []
for l in leaves:
if l.data.get(kmer):
x.append(l)
return set(x)
for kmer in kmers:
assert set(root.find(search_kmer, kmer)) == search_kmer_in_list(kmer)
print('-----')
print([x.metadata for x in root.find(search_kmer, "AAAAA")])
print([x.metadata for x in root.find(search_kmer, "AAAAT")])
print([x.metadata for x in root.find(search_kmer, "AAAAG")])
print([x.metadata for x in root.find(search_kmer, "CAAAA")])
print([x.metadata for x in root.find(search_kmer, "GAAAA")])
with utils.TempDirectory() as location:
root.save(os.path.join(location, 'demo'))
root = SBT.load(os.path.join(location, 'demo'))
for kmer in kmers:
new_result = {str(r) for r in root.find(search_kmer, kmer)}
print(*new_result, sep='\n')
assert new_result == {str(r) for r in search_kmer_in_list(kmer)}