def test_scores_bad_input(self):
"""inference function should raise with bad input."""
# model is empty
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
with self.assertRaises(Exception):
rs.inference("refA")
def test_add_without_name_raises(self):
"""adding a representation without a name raises exceptions."""
r1 = Representation().set("DOID:0014667", 0.4)
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
with self.assertRaises(Exception):
rs.add(r1)
def setUp(self):
"""prepare a reference set with some data."""
# let ref universe have two annotations and one null
refA = Representation(data=dict(a=1, b=0.8), name="refA")
refA.defaults(zerovals)
refB = Representation(data=dict(a=1, d=0.2), name="refB")
refB.defaults(zerovals)
self.rs = ReferenceSet(dict(null=0.7, refA=0.15, refB=0.15),
ids=zerovals.keys())
self.rs.add(refA).add(refB)
self.rs.prep()
def dict2referenceset(repdict, feature_ids, priors):
"""create a representation set, using imputation
:param repdict: a dictionary of Representation objects
:param feature_ids: list with all features
:param priors: dict linking name to a prior probability
:return: ReferenceSet object
"""
result = ReferenceSet(priors, feature_ids)
for id, representation in repdict.items():
result.add(representation)
return result
def test_get_reference(self):
"""extract one reference from a representation set."""
r1 = Representation(name="refA").set("DOID:0014667", 0.4)
r1.impute(obo, obodefaults)
r2 = Representation(name="refB").set("DOID:0080015", 0.6)
r2.impute(obo, obodefaults)
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
rs.add(r1).add(r2)
r3 = rs.get_representation("refA")
self.assertTrue(r3.equal(r1))
def test_positive_parent_multi(self):
"""fetching a parent term when terms have multiple parents."""
# load an ontology in which Y7 is connected to both Y2 and Y1
Yfile = join(testdir, "Ymulti.obo")
Yobo = MinimalObo(Yfile)
Ydefaults = dict.fromkeys(Yobo.ids(), 0.0001)
Ydefaults["Y:003"] = 0.0002
Ydefaults["Y:005"] = 0.0002
# make slight variations of representations
rs = ReferenceSet(dict(refA=0.5, refB=0.5),
ids=Yobo.ids(),
row_priors=Ydefaults)
refA = Representation(name="refA")
refA.set("Y:002", 0.5).set("Y:005", 1).impute(Yobo, Ydefaults)
refB = Representation(name="refB")
refB.set("Y:001", 0.5).impute(Yobo, Ydefaults)
rs.add(refA).add(refB)
rs.learn_obo(Yobo)
self.assertEqual(
rs._positive_ancestor(rs.columns["refA"], rs.rows["Y:007"]),
rs.rows["Y:002"], "Y2 is a positive ancestor")
self.assertEqual(
rs._positive_ancestor(rs.columns["refB"], rs.rows["Y:007"]),
rs.rows["Y:001"], "Y1 is a positive immediate parent")
def test_between2(self):
"""inference when model equally similar to two refs"""
# let ref universe have two annotations
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=self.obo.ids(),
row_priors=self.obodefaults)
rs.add(self.refA).add(self.refB)
rs.learn_obo(self.obo)
rs.prep()
inf = rs.inference(self.y3model)
self.assertAlmostEqual(inf["refA"], inf["refB"], msg="equally likely")
def setUpClass(cls):
# create with sibling diseases Y:002 and Y:001 are siblings
# set some phenotype priors that are nonzero
cls.priors = dict()
cls.priors["Y:004"] = 0.66
cls.priors["Y:005"] = cls.priors["Y:006"] = 0.25
cls.priors["Y:003"] = 0.66
cls.priors["Y:001"] = cls.priors["Y:002"] = 0.33
cls.priors["Y:007"] = cls.priors["Y:008"] = 0.25
cls.refnull = Representation(name="null")
# refA has a negative phenotype
cls.refA = Representation(name="refA")
cls.refA.set("Y:002", 0.1).impute(Yobo, cls.priors)
# refB has a negative and positive phenotypes
cls.refB = Representation(name="refB")
cls.refB.set("Y:001", 0.01).set("Y:006", 0.8).impute(Yobo, cls.priors)
# refB2 has a weaker positive phenotype
cls.refB2 = Representation(name="refB2")
cls.refB2.set("Y:001", 0.1).set("Y:006", 0.5).impute(Yobo, cls.priors)
cls.rs = ReferenceSet(dict(null=0.4, refA=0.3, refB=0.3, refB2=0.3),
ids=Yobo.ids(),
row_priors=cls.priors)
cls.rs.add(cls.refnull).add(cls.refA).add(cls.refB).add(cls.refB2)
cls.rs.learn_obo(Yobo)
def setUpClass(cls):
# set some phenotype priors that are nonzero
cls.priors = dict()
cls.priors["Y:004"] = 0.66
cls.priors["Y:005"] = cls.priors["Y:006"] = 0.25
cls.priors["Y:003"] = 0.66
cls.priors["Y:001"] = cls.priors["Y:002"] = 0.33
cls.priors["Y:007"] = cls.priors["Y:008"] = 0.25
# create reference set with some strong phenotypes
cls.refnull = Representation(name="null")
cls.refA = Representation(name="refA")
cls.refB = Representation(name="refB")
cls.refA.set("Y:002", 1).impute(Yobo, cls.priors)
cls.refB = Representation(name="refB")
cls.refB.set("Y:001", 1).impute(Yobo, cls.priors)
# reset missing phenotypes to smaller-than-prior
for k, v in cls.priors.items():
if cls.refA.get(k) == v:
cls.refA.set(k, v / 2)
if cls.refB.get(k) == v:
cls.refB.set(k, v / 2)
cls.rs = ReferenceSet(dict(null=0.3, refA=0.3, refB=0.3),
ids=Yobo.ids(),
row_priors=cls.priors)
cls.rs.add(cls.refnull).add(cls.refA).add(cls.refB)
cls.rs.learn_obo(Yobo)
def test_between_refs_and_null(self):
"""inference when model is similar to two refs and a there is a null"""
# let ref universe have two annotations and one null
rs = ReferenceSet(dict(null=0.8, refA=0.15, refB=0.15),
ids=self.obo.ids(), row_priors=self.obodefaults)
rs.add(self.refA).add(self.refB)
rs.learn_obo(self.obo)
rs.prep()
inf = rs.inference(self.y3model)
self.assertAlmostEqual(inf["refA"], inf["refB"],
msg="equally likely")
def test_empty_representation(self):
"""creating a new set of references."""
rs = ReferenceSet(dict(refA=0.5, refB=1), ids=obo.ids())
num_ids = len(obo.ids())
self.assertEqual(len(rs.data), 2, "refset should allocate memory")
self.assertEqual(len(rs.data[0]), num_ids)
self.assertEqual(len(rs.data[1]), num_ids)
def test_str(self):
"""getting a quick string with the content."""
rs = ReferenceSet(dict(null=0.7, refA=0.15, refB=0.15),
ids=zerovals.keys())
result = str(rs)
self.assertTrue("refA" in result)
self.assertFalse("hello" in result)
def test_difference_in_priors(self):
"""inference when model matches two references,
but have different priors"""
# let ref universe have two annotations and one null
rs = ReferenceSet(dict(null=0.85, refA=0.05, refB=0.1),
ids=self.obo.ids(), row_priors=self.obodefaults)
rs.add(self.refA).add(self.refB)
rs.learn_obo(self.obo)
rs.prep()
inf = rs.inference(self.y3model)
self.assertLess(inf["refA"], inf["refB"],
msg="equal match, but A has weaker prior")
def test_model_nodata(self):
"""inference when references are unequal but model has no data"""
model = Representation(name="nodata")
# let ref universe have two annotations and one null
rs = ReferenceSet(dict(null=0.8, refA=0.1, refB=0.1),
ids=self.obo.ids(), row_priors=self.obodefaults)
rs.add(self.refA).add(self.refB)
rs.learn_obo(self.obo)
rs.prep()
inf = rs.inference(model)
self.assertAlmostEqual(inf["refA"], inf["refB"],
msg="equally likely")
def test_positive_parent(self):
"""fetching a parent term that has a positive value."""
rs = ReferenceSet(dict(refA=0.5), ids=Yobo.ids(), row_priors=Ydefaults)
refA = Representation(name="refA")
refA.set("Y:002", 0.5).set("Y:005", 1).impute(Yobo, Ydefaults)
rs.add(refA)
rs.learn_obo(Yobo)
refAindex = rs.columns["refA"]
self.assertEqual(rs._positive_ancestor(refAindex, rs.rows["Y:002"]),
rs.rows["Y:002"], "Y2 is itself is positive")
self.assertEqual(rs._positive_ancestor(refAindex, rs.rows["Y:007"]),
rs.rows["Y:002"], "Y2 is immediate parent of Y7")
self.assertEqual(rs._positive_ancestor(refAindex, rs.rows["Y:006"]),
rs.rows["Y:005"], "Y5 is immediate parent of Y6")
def test_baddata(self):
"""inference should raise when input is bad"""
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=["a", "b", "c"])
rs.prep()
with self.assertRaises(Exception) as e:
rs.inference(5)
def test_prep_row_priors(self):
"""prepare row priors."""
# let ref universe have two annotations and one null
refA = Representation(data=dict(a=1), name="refA")
refA.defaults(zerovals)
refB = Representation(data=dict(a=1, b=0.8), name="refB")
refB.defaults(zerovals)
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=zerovals.keys())
rs.add(refA).add(refB)
# compute feature priors
rs.prep()
# row_priors should gain key/values for all features
expected_features = set(zerovals.keys())
self.assertEqual(set(rs.row_names), expected_features)
# features declared in representations should get reasonable priors
a_index = rs.rows["a"]
b_index = rs.rows["b"]
d_index = rs.rows["d"]
self.assertEqual(rs.row_priors[a_index], 1,
"refA and refB both have a")
self.assertEqual(rs.row_priors[b_index], 0.4,
"only refB has b, so 0.8/2")
self.assertEqual(rs.row_priors[d_index], 0.2,
"value is 1/num features")
def test_learn_from_obo(self):
"""create parents_of tuples for all features"""
r1 = Representation(name="refA").set("DOID:0014667", 0.4)
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
rs.add(r1)
self.assertEqual(rs.parents, None)
rs.learn_obo(obo)
self.assertEqual(len(rs.parents), len(obo.ids()))
def test_subset(self):
"""make a refset smaller by ignoring some features."""
# create a reference set
rs = ReferenceSet(OrderedDict(refA=0.5, refB=0.5),
ids=testfeatures,
row_priors=zerovals)
# add some data to the two representations
r1 = Representation(name="refA")
r1.set("a", 0.1).set("b", 0.2).set("c", 0.3).set("d", 0.4)
r2 = Representation(name="refB")
r2.set("c", 0.6).set("d", 0.7).set("e", 0.8)
rs.add(r1).add(r2)
# manually create arrays with the reference set data
expected_raw_A = [0.1, 0.2, 0.3, 0.4, 0.0]
expected_raw_B = [0.0, 0.0, 0.6, 0.7, 0.8]
self.assertTrue(list(rs.data[0]), expected_raw_A)
self.assertTrue(list(rs.data[1]), expected_raw_B)
# subset to a smaller number of features
# myids - here c is repeated twice, z is not in the original features
myids = ["e", "c", "a", "z", "c"]
rs = ReferenceMatrix(rs, myids)
# check new shape (three features and two references)
self.assertEqual(len(rs.rows), 3)
self.assertEqual(len(rs.row_names), 3)
self.assertEqual(rs.data.shape, (3, 2))
# check that the relevant rows are present
result = set(rs.rows.keys())
expected = set(myids)
expected.remove("z")
self.assertEqual(result, expected)
# check data subset in output
output_A = [0.1, 0.3, 0.0]
output_B = [0.0, 0.6, 0.8]
self.assertEqual(sum(rs.data[:, 0]), sum(output_A))
self.assertEqual(sum(rs.data[:, 1]), sum(output_B))
def get_modelsets(dbpath, obo, partition_size=4096):
"""create ReferenceSets objects with general and specific phenotypes
:param dbpath: path to phenoscoring db
:param config: dictionary configuration settings
:param obo: object with ontology
:return: array of ReferenceSets objects, each with a subset of models
"""
model_names = get_model_names(dbpath)
if len(model_names) == 0:
return []
# partition models into chunks
model_groups = [[]]
for m in model_names:
group = model_groups[-1]
if len(group) >= partition_size:
model_groups.append([])
group = model_groups[-1]
group.append(m)
# load all model information from database
models = get_model_representations(dbpath, obo)
phen_priors = get_phenotype_priors(dbpath)
# transfer into small-sized reference sets
result = []
for group in model_groups:
packet_priors = dict.fromkeys(group, 1 / len(model_names))
refset = ReferenceSet(packet_priors, obo.ids())
for m in group:
model = models.pop(m)
model.impute(obo, phen_priors)
refset.add(model)
result.append(refset)
return result
def test_add_raises(self):
"""adding an unexpected piece of data raises exceptions."""
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
with self.assertRaises(Exception):
rs.add(5)
with self.assertRaises(Exception):
rs.add(dict.fromkeys(["DOID:0014667", "DOID:0080015"], 0))
def get_refsets(dbpath, ref_priors=None, phenotype_priors=None):
"""create ReferenceSets objects with general and specific phenotypes
:param dbpath: path to phenoscoring db
:param ref_priors: dictionary with priors for references
(if None, fetched from db)
:param phenotype_priors: dictionary with priors for all featurs
(if None, fetched from db)
:return: two ReferenceSets objects
"""
# at first create just a dictionary of representations
general_dict, specific_dict = dict(), dict()
if phenotype_priors is None:
phenotype_priors = get_phenotype_priors(dbpath)
if ref_priors is None:
ref_priors = get_ref_priors(dbpath)
# create empty Representations for each reference
nullrep = get_complete_null(dbpath)
phenotypes = nullrep.keys()
for id in ref_priors.keys():
general_dict[id] = nullrep.copy(name=id)
specific_dict[id] = nullrep.copy(name=id)
# fill the representations with values
phentab = ReferenceCompletePhenotypeTable(dbpath)
if len(ref_priors) == 1:
refname = list(ref_priors.keys())[0]
generator = DBGenerator(phentab, where=dict(id=refname))
else:
generator = DBGenerator(phentab)
for row in generator.next():
id, phen = row["id"], row["phenotype"]
if id in ref_priors:
general_dict[id].set(phen, row["value"])
specific_dict[id].set(phen, row["specific_value"])
# transfer representations into ReferenceSets
general = ReferenceSet(ref_priors, phenotypes, phenotype_priors)
specific = ReferenceSet(ref_priors, phenotypes, phenotype_priors)
for refid in general_dict.keys():
general.add(general_dict[refid])
specific.add(specific_dict[refid])
return general, specific
def test_underflow(self):
"""attempt to get underflow in individual p."""
# let model have very sure values
model = Representation(name="underflow")
model.set("Y:007", 0.00001).set("Y:004", 1).set("Y:003", 1)
# let ref universe have two annotations and one null
refA = Representation(name="refA").set("Y:003", 1)
refB = Representation(name="refB").set("Y:003", 1)
rs = ReferenceSet(dict(null=0.98, refA=0.001, refB=0.001),
ids=self.obo.ids())
rs.add(refA).add(refB)
rs.learn_obo(self.obo)
rs.prep()
result = rs.inference(model, verbose=True)
self.assertGreaterEqual(result["refA"], 0,
msg="must always be a number, even if zero")
self.assertGreaterEqual(result["refB"], 0,
msg="must always be a number, even if zero")
self.assertGreaterEqual(result["refB"], 0,
msg="must always be a number, even if zero")
def test_FP_can_increase(self):
"""FP can in principle yield greater score"""
# make a new reference set with different priors
priors2 = self.priors.copy()
priors2["Y:002"] = 0.1
rs2 = ReferenceSet(dict(null=0.4, refA=0.3, refB=0.3),
ids=Yobo.ids(),
row_priors=priors2)
rs2.add(self.refnull).add(self.refA).add(self.refB)
rs2.learn_obo(Yobo)
FP = Representation(name="model").set("Y:002", 0.2)
chain = rs2.inference_chain(FP, "refB", verbose=True, fp_penalty=1)
chain.evaluate_inference()
self.assertGreater(chain.posterior, chain.prior)
def test_add_incrementally(self):
"""transferring values into a representation set."""
r1 = Representation(name="refA").set("DOID:0014667", 0.4)
r1.impute(obo, obodefaults)
r2 = Representation(name="refB").set("DOID:0080015", 0.6)
r2.impute(obo, obodefaults)
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=obo.ids())
rs.add(r1).add(r2)
self.assertEqual(rs.get("DOID:0014667", "refA"), 0.4,
"refset should contain inserted data")
self.assertEqual(rs.get("DOID:0080015", "refB"), 0.6,
"refset should contain inserted data")
self.assertEqual(rs.get("DOID:4", "refB"), 0.6,
"refset should contain imputed data")
def test_TP_scores_better_than_FP(self):
"""FPs must score lower than TPs"""
# make a new reference set with different (lower) priors
priors2 = self.priors.copy()
priors2["Y:002"] = 0.55
priors2["Y:007"] = 0.15
rs2 = ReferenceSet(dict(null=0.4, refA=0.3, refB=0.3),
ids=Yobo.ids(),
row_priors=priors2)
rs2.add(self.refnull).add(self.refA).add(self.refB)
rs2.learn_obo(Yobo)
# compare with refA, which has Y:002 equal to 1
FP = Representation(name="FP").set("Y:007", 1)
chain_FP = rs2.inference_chain(FP, "refA", verbose=True, fp_penalty=2)
chain_FP.evaluate_inference()
TP = Representation(name="TP").set("Y:002", 1)
chain_TP = rs2.inference_chain(TP, "refA", verbose=True)
chain_TP.evaluate_inference()
self.assertGreaterEqual(chain_TP.posterior, chain_FP.posterior)
def test_FP_with_fp_penalty(self):
"""FP increases more with lower fp_penalty"""
# make a new reference set with different priors
priors2 = self.priors.copy()
priors2["Y:003"] = 0.4
priors2["Y:002"] = 0.15
priors2["Y:007"] = 0.1
rs2 = ReferenceSet(dict(null=0.4, ref=0.3),
ids=Yobo.ids(),
row_priors=priors2)
ref = Representation(name="ref")
ref.set("Y:001", 1).impute(Yobo, priors2)
ref.set("Y:007", priors2["Y:007"] / 2)
rs2.add(self.refnull).add(ref)
rs2.learn_obo(Yobo)
FP = Representation(name="model").set("Y:007", 0.35)
chain1 = rs2.inference_chain(FP, "ref", verbose=True, fp_penalty=0.1)
chain1.evaluate_inference()
self.assertGreater(chain1.posterior, chain1.prior)
chain2 = rs2.inference_chain(FP, "ref", verbose=True, fp_penalty=1)
chain2.evaluate_inference()
self.assertLess(chain2.posterior, chain1.posterior)
def setUp(self):
"""prepare a reference set with a few references."""
self.refnull = Representation(data=null_defaults, name="null")
self.refA = Representation(data=Ydefaults, name="refA")
self.refA.set("Y:004", 1)
self.refB = Representation(data=Ydefaults, name="refB")
self.refB.set("Y:004", 1).set("Y:001", 0.6).set("Y:003", 0.5)
self.refC = Representation(data=Ydefaults, name="refC")
self.refC.set("Y:004", 1).set("Y:001", 0.5)
self.refD = Representation(data=Ydefaults, name="refD")
self.refD.set("Y:004", 0.1)
ref_priors = dict(null=0.1, refA=0.1, refB=0.1, refC=0.1, refD=0.1)
rs = ReferenceSet(ref_priors, ids=Ydefaults.keys())
rs.add(self.refnull)
rs.add(self.refA).add(self.refB)
rs.add(self.refC).add(self.refD)
self.rm = ReferenceMatrix(rs, list(Ydefaults.keys()))
def test_inference_chain(self):
"""compute an inference chain."""
# create a reference set
refA = Representation(name="refA")
refA.set("Y:002", 1).impute(Yobo, Ydefaults)
refB = Representation(name="refB")
refB.set("Y:001", 1).impute(Yobo, Ydefaults)
rs = ReferenceSet(dict(refA=0.5, refB=0.5),
ids=Yobo.ids(),
row_priors=Ydefaults)
rs.add(refA).add(refB)
rs.learn_obo(Yobo)
# compute a chain object explaining scoring steps
chain = rs.inference_chain(refA, "refB", verbose=True)
self.assertEqual(chain.__dict__["model"], "refA")
self.assertEqual(chain.__dict__["reference"], "refB")
self.assertGreater(len(chain.data), 2,
"data chain should describe multiple features")
self.assertTrue("background" in chain.data[0].__dict__,
"chain data have comparison information")
self.assertTrue("result" in chain.data[0].__dict__,
"chain data have TP/FP/etc codes")
def test_checkprep(self):
"""inference only works when set is prepped"""
rs = ReferenceSet(dict(refA=0.5, refB=0.5), ids=self.obo.ids())
with self.assertRaises(Exception):
rs.inferenceModel(self.y3model)