def test_count_lca_for_assignments_abund_3():
# test basic mechanics of gather_assignments function with two lineages
# and two hashvals
hashval = 12345678
hashval2 = 87654321
hashval_counts = dict()
hashval_counts[hashval] = 2
hashval_counts[hashval2] = 5
lin = lca_utils.make_lineage('a;b;c')
lin2 = lca_utils.make_lineage('a;b;d')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin, lin2 ]))
db._set_lineage_assignment(hashval2, set([ lin ]))
assignments = lca_utils.gather_assignments(hashval_counts, [db])
counts = count_lca_for_assignments(assignments, hashval_counts)
print(counts)
assert len(counts) == 2
assert counts[lin] == 5 # makes sense
assert counts[lin2] == 0 # makes sense
lca_lin = lca_utils.make_lineage('a;b')
assert counts[lca_lin] == 2 # yes!
def test_count_lca_for_assignments_abund_4():
# test basic mechanics of gather_assignments function with three lineages
# and three hashvals
hashval = 12345678
hashval2 = 87654321
hashval3 = 34567891
hashval_counts = dict()
hashval_counts[hashval] = 2
hashval_counts[hashval2] = 5
hashval_counts[hashval3] = 3
lin = lca_utils.make_lineage('a;b;c')
lin2 = lca_utils.make_lineage('a;b;d')
lin3 = lca_utils.make_lineage('a;b;d;e')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin, lin2 ])) # lca: a;b
db._set_lineage_assignment(hashval2, set([ lin ])) # lca: a;b;c
db._set_lineage_assignment(hashval3, set([ lin2, lin3 ])) # a;b;d;e
assignments = lca_utils.gather_assignments(hashval_counts, [db])
counts = count_lca_for_assignments(assignments, hashval_counts)
print(counts)
assert len(counts) == 3
assert counts[lin] == 5 # makes sense b/c hashval2
assert counts[lin2] == 0 # a;b;d (lin2) + a;b;d;e (lin3) -->a;b;d;e (lin3) only
assert counts[lin3] == 3 # hashval3
lca_lin = lca_utils.make_lineage('a;b')
assert counts[lca_lin] == 2 # yes, b/c hashval
def summarize(hashvals, dblist, threshold):
"""
Classify 'hashvals' using the given list of databases.
Insist on at least 'threshold' counts of a given lineage before taking
it seriously.
Return (lineage, counts) where 'lineage' is a tuple of LineagePairs.
"""
# gather assignments from across all the databases
assignments = lca_utils.gather_assignments(hashvals, dblist)
# now convert to trees -> do LCA & counts
counts = lca_utils.count_lca_for_assignments(assignments)
# ok, we now have the LCAs for each hashval, and their number
# of counts. Now aggregate counts across the tree, going up from
# the leaves.
aggregated_counts = defaultdict(int)
for lca, count in counts.most_common():
if count < threshold:
break
if not lca:
aggregated_counts[lca] += count
# climb from the lca to the root.
aggregated_counts[lca] += count
return aggregated_counts
def test_gather_assignments_1():
# test basic mechanics of gather_assignments function
hashval = 12345678
lin = lca_utils.make_lineage('a;b;c')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin ]))
assignments = lca_utils.gather_assignments([hashval], [db])
print(assignments)
assert assignments[hashval] == set([ lin ])
def test_count_lca_for_assignments_abund_1():
# test basic mechanics of gather_assignments function
hashval = 12345678
hashval_counts = dict()
hashval_counts[hashval] = 3
lin = lca_utils.make_lineage('a;b;c')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin ]))
assignments = lca_utils.gather_assignments(hashval_counts.keys(), [db])
counts = count_lca_for_assignments(assignments, hashval_counts)
print(counts)
assert len(counts) == 1
assert counts[lin] == 3
def test_gather_assignments_3():
# test basic mechanics of gather_assignments function with two lineages
# and two hashvals
hashval = 12345678
hashval2 = 87654321
lin = lca_utils.make_lineage('a;b;c')
lin2 = lca_utils.make_lineage('a;b;d')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin, lin2 ]))
db._set_lineage_assignment(hashval2, set([ lin ]))
assignments = lca_utils.gather_assignments([hashval, hashval2], [db])
print(assignments)
assert assignments[hashval] == set([ lin, lin2 ])
assert assignments[hashval2] == set([ lin ])
def test_count_lca_for_assignments_2():
# test basic mechanics of gather_assignments function with two lineages
hashval = 12345678
lin = lca_utils.make_lineage('a;b;c')
lin2 = lca_utils.make_lineage('a;b;d')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin, lin2 ]))
assignments = lca_utils.gather_assignments([hashval], [db])
counts = count_lca_for_assignments(assignments)
print(counts)
assert counts[lin] == 0
assert counts[lin2] == 0
assert len(counts) == 1
lca_lin = lca_utils.make_lineage('a;b')
assert counts[lca_lin] == 1
def test_count_lca_for_assignments_abund_5():
# test basic mechanics of gather_assignments function with two lineages
# and two hashvals when linages match but one has lower taxo detail
hashval = 12345678
hashval2 = 87654321
hashval_counts = dict()
hashval_counts[hashval] = 2
hashval_counts[hashval2] = 5
lin = lca_utils.make_lineage('a;b;d')
lin2 = lca_utils.make_lineage('a;b;d;e')
db = FakeLCA_Database()
db._set_lineage_assignment(hashval, set([ lin, lin2 ]))
db._set_lineage_assignment(hashval2, set([ lin ]))
assignments = lca_utils.gather_assignments(hashval_counts, [db])
counts = count_lca_for_assignments(assignments, hashval_counts)
print(counts)
assert len(counts) == 2
assert counts[lin] == 5 # makes sense
assert counts[lin2] == 2 # lin+lin2 yield just lin2