Merge pull request #3 from opentargets/ag_efo_latest_obo

Ag efo latest obo, update therapeutic areas and remove splitting query strings

CHANGES

@@ -1,3 +1,8 @@
+Version 0.0.14:
+ - Query strings are searched as they are, splitting at commas and semicolons and using the largest substring has been removed.
+ - Use latest EFO OBO file instead of hard-coded v2018-01-15
+ - Updated OT top nodes with current therapeutic areas for checking if mapped term is in OT
+
 Version 0.0.13:
  - even more weird and wonderful packaging problems
 

Pipfile

@@ -14,9 +14,9 @@ recommonmark = "*"
 sphinx-rtd-theme = "*"
 
 [packages]
-requests = "==2.19.1"
+requests = "*"
 obonet = "*"
 click = "*"
 
 [requires]
-python_version = "3.6"
+python_version = "3.7"

VERSION

@@ -1 +1 @@
-0.0.13
+0.0.14

ontoma/constants.py

@@ -2,21 +2,62 @@
 
 URLS = {
 
-    'EFO':'https://github.com/EBISPOT/efo/raw/v2018-01-15/efo.obo',
+    'EFO':'https://www.ebi.ac.uk/efo/efo.obo',
     'HP':'http://purl.obolibrary.org/obo/hp.obo',
     'OMIM_EFO_MAP':'https://raw.githubusercontent.com/opentargets/platform_semantic/master/resources/xref_mappings/omim_to_efo.txt',
     'ZOOMA_EFO_MAP':'https://raw.githubusercontent.com/opentargets/platform_semantic/master/resources/zooma/cttv_indications_3.txt',
     'MONDO':'http://purl.obolibrary.org/obo/mondo.obo'
 }
 OT_TOP_NODES = {
-    'http://www.ebi.ac.uk/efo/EFO_0000408',
-    'http://www.ebi.ac.uk/efo/EFO_0000651',
-    'http://www.ebi.ac.uk/efo/EFO_0001444',
-    'http://purl.obolibrary.org/obo/GO_0008150',
-    'http://www.ifomis.org/bfo/1.1/snap#Function',
-    'http://www.ebi.ac.uk/efo/EFO_0000546',
-    'http://www.ebi.ac.uk/efo/EFO_0003935',
-    'http://purl.obolibrary.org/obo/HP_0000118'
+    'http://www.ebi.ac.uk/efo/EFO_0005932', # Animal disease
+    'http://purl.obolibrary.org/obo/GO_0008150', # Biological process
+    'http://purl.obolibrary.org/obo/MONDO_0045024', # Cell proliferation disorder
+    'http://purl.obolibrary.org/obo/MONDO_0021205', # Disease of ear
+    'http://purl.obolibrary.org/obo/MONDO_0024458', # Disease of visual system
+    'http://www.ebi.ac.uk/efo/EFO_0001379', # Endocrine system disease
+    'http://www.ebi.ac.uk/efo/EFO_0010282', # Gastrointestinal disease
+    'http://www.ebi.ac.uk/efo/OTAR_0000018', # Genetic, familial or congenital disease
+    'http://www.ebi.ac.uk/efo/EFO_0000508', # Genetic disorder (child of OTAR_0000018)
+    'http://www.ebi.ac.uk/efo/OTAR_0000019', # Familial disease (child of OTAR_0000018, does not have any children)
+    'http://www.ebi.ac.uk/efo/MONDO_0018797', # Genetic cardiac malformation (child of OTAR_0000018)
+    'http://www.ebi.ac.uk/efo/MONDO_0000839', # Congenital abnormality (child of OTAR_0000018)
+    'http://www.ebi.ac.uk/efo/EFO_0000319', # Cardiovascular disease
+    'http://www.ebi.ac.uk/efo/EFO_0005803', # Hematologic disease
+    'http://www.ebi.ac.uk/efo/EFO_0000540', # Immune system disease
+    'http://www.ebi.ac.uk/efo/EFO_0005741', # Infectious disease
+    'http://www.ebi.ac.uk/efo/OTAR_0000009', # Injury, poisoning or other complication
+    'http://www.ebi.ac.uk/efo/EFO_0008546' , # Poisoning (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_1000903' , # Drug-induced akathisia (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_1000904' , # Drug-Induced dyskinesia (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/MONDO_0001423' , # Drug-induced mental disorder (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0004228' , # Drug-induced liver injury (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0009482' , # Drug allergy  (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0009518' , # Complication (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0000546' , # Injury (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0003099' , # Cushing syndrome (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/MONDO_0016474' , # Drug-induced lupus erythematosus (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0005400' , # chemotherapy-induced alopecia (child of OTAR_0000009)
+    'http://www.ebi.ac.uk/efo/EFO_0010285', # Integumentary system disease
+    'http://www.ebi.ac.uk/efo/EFO_0001444', # Measurement
+    'http://www.ebi.ac.uk/efo/OTAR_0000006', # Musculoskeletal or connective tissue disease
+    'http://www.ebi.ac.uk/efo/EFO_1001986', # Connective tissue disease (child of OTAR_0000006)
+    'http://www.ebi.ac.uk/efo/EFO_0009676', # Musculoskeletal system disease (child of OTAR_0000006)
+    'http://www.ebi.ac.uk/efo/EFO_0000618', # Nervous system disease
+    'http://purl.obolibrary.org/obo/MONDO_0024297', # Nutritional or metabolic disease
+    'http://www.ebi.ac.uk/efo/EFO_0009605', # Pancreas disease
+    'http://www.ebi.ac.uk/efo/EFO_0000651', # Phenotype
+    'http://www.ebi.ac.uk/efo/OTAR_0000014', # Pregnancy or perinatal disease
+    'http://www.ebi.ac.uk/efo/EFO_0009683', # Puerperal disorder (child of OTAR_0000014)
+    'http://www.ebi.ac.uk/efo/EFO_0009682', # Pregnancy disorder (child of OTAR_0000014)
+    'http://www.ebi.ac.uk/efo/EFO_0010238', # Perinatal disease (child of OTAR_0000014)
+    'http://purl.obolibrary.org/obo/MONDO_0002025', # Psychiatric disorder
+    'http://www.ebi.ac.uk/efo/OTAR_0000017', # Reproductive system or breast disease
+    'http://www.ebi.ac.uk/efo/EFO_0000512', # Reproductive system disease (child of OTAR_0000017)
+    'http://www.ebi.ac.uk/efo/EFO_0009483', # Breast disease (child of OTAR_0000017)
+    'http://www.ebi.ac.uk/efo/OTAR_0000010', # Respiratory or thoracic disease
+    'http://www.ebi.ac.uk/efo/EFO_0000684', # Respiratory system disease (child of OTAR_0000010)
+    'http://www.ebi.ac.uk/efo/MONDO_0000651', # Thoracic disease (child of OTAR_0000010)
+    'http://www.ebi.ac.uk/efo/EFO_0009690', # Urinary system disease
 }
 
 FIELDS = ['query','term','label','source','quality','action']

ontoma/downloaders.py

@@ -19,7 +19,7 @@ def get_omim_to_efo_mappings(url):
     '''returns a dictionary that maps OMIM codes to EFO_uri
     >>> d = get_omim_to_efo_mappings(URLS['OMIM_EFO_MAP'])
     >>> d['609909']
-    [{'iri': 'http://www.orpha.net/ORDO/Orphanet_154', 'label': 'Familial isolated dilated cardiomyopathy'}, {'iri': 'http://www.orpha.net/ORDO/Orphanet_217607', 'label': 'Familial dilated cardiomyopathy'}]
+    [{'iri': 'http://www.orpha.net/ORDO/Orphanet_217607', 'label': 'Familial dilated cardiomyopathy'}, {'iri': 'http://www.orpha.net/ORDO/Orphanet_154', 'label': 'Familial isolated dilated cardiomyopathy'}]
     '''
     mappings = {}
     logger.debug("OMIM to EFO mappings - requesting from URL %s", url)

ontoma/interface.py

@@ -110,10 +110,6 @@ def make_uri(ontology_short_form):
                          "Short form: {} not recognized".format(ontology_code))
 
 
-def largest_fragment(sentence):
-    return max(re.split("[,;]+",sentence),key=len).strip()
-
-
 class OnToma(object):
     '''Open Targets ontology mapping wrapper
 
@@ -159,7 +155,7 @@ class OnToma(object):
         OMIM code lookup
 
         >>> t.omim_lookup('230650')
-        'http://www.orpha.net/ORDO/Orphanet_354'
+        'http://www.orpha.net/ORDO/Orphanet_79257'
 
         >>> t.zooma_lookup('asthma')
         'http://www.ebi.ac.uk/efo/EFO_0000270'
@@ -168,11 +164,6 @@ class OnToma(object):
         >>> t.mondo_lookup('asthma')
         'http://purl.obolibrary.org/obo/MONDO_0004979'
 
-        Searching the ICD9 code for 'other dermatoses' returns EFO's skin disease:
-
-        >>> t.icd9_lookup('696')
-        'EFO:0000676'
-
         There is also a semi-intelligent wrapper, which tries to guess the
         best matching strategy:
 
@@ -295,6 +286,7 @@ def otzooma_map_lookup(self, name):
 
     def icd9_lookup(self, icd9code):
         '''Searches the ICD9CM <=> EFO mappings returned from the OXO API
+        #FIXME Results don't seem to be deterministic, some mappings appear and disappear between calls, e.g. t.icd9_lookup('696')
         '''
         return self._icd9_to_efo[icd9code]
 
@@ -348,11 +340,8 @@ def _is_included(self, iri, ontology=None):
         we select for open targets
         '''
         if not ontology:
-            if 'HP_' in iri:
-                ontology = 'hp'
-            else:
-                # default to checking ancestry in EFO
-                ontology = 'efo'
+            # default to checking ancestry in EFO
+            ontology = 'efo'
         try:
             for ancestor in self._ols.get_ancestors(ontology, iri):
                 if ancestor['iri'] in OT_TOP_NODES:
@@ -421,7 +410,7 @@ def find_term(self, query, code=None, suggest=False, verbose=False):
                             'for %s in %s mappings. ', e, code)
                 return None
         else:
-            found = self._find_term_from_string(largest_fragment(query), suggest)
+            found = self._find_term_from_string(query, suggest)
             if found:
                 msg = 'Found {} for {} from {} - {} - {}'.format(
                             make_uri(found['term']),

ontoma/ols.py

@@ -72,8 +72,8 @@ class OlsClient:
     >>> ols.besthit('hypogammaglobulinemia',ontology='efo')['label']
     'Osteopetrosis - hypogammaglobulinemia'
 
-    >>> ols.besthit('hypogammaglobulinemia',ontology='efo',exact=True) is None
-    True
+    >>> ols.besthit('hypogammaglobulinemia',ontology='efo',exact=True)['label']
+    'Agammaglobulinemia'
 
     >>> r = ols.search('asthma',ontology=['efo'],query_fields=['synonym'],field_list=['iri','label'])
     >>> 'http://www.ebi.ac.uk/efo/EFO_0004591' in [syn['iri'] for syn in r]
@@ -98,7 +98,7 @@ class OlsClient:
     'EFO_1001054'
 
     >>> [x['short_form'] for x in ols.select('alzheimer')[:2]]
-    ['NCIT_C2866', 'NCIT_C38778']
+    ['PW_0000015', 'DOID_0080348']
 
     You can also pass your favourite parameters at class instantiation:
 

ontoma/oxo.py

@@ -25,22 +25,7 @@
 
 class OxoClient:
     ''' OXO wrapper class
-
-    >>> oxo = OxoClient()
-    >>> len(oxo._sources)
-    940
-
-    >>> first_result = list(oxo.search(input_source="ICD9CM"))[:1][0]
-    >>> first_result['curie']
-    'ICD9CM:730.92'
-
-    >>> for r in oxo.search(ids=['ICD9CM:171.6'],input_source="ICD9CM"):
-    ...     print(r['label'])
-    Malignant neoplasm of connective and other soft tissue of pelvis
-
-    >>> icd9s = oxo.make_mappings(input_source="ICD9CM", distance=2)
-    >>> icd9s['733.0']
-    'EFO:0003882'
+    FIXME Results don't seem to be deterministic, some mappings appear and disappear between calls, e.g. icd9s = oxo.make_mappings(input_source="ICD9CM", distance=2); icd9s['733.09']
     '''
 
     def __init__(self, base_url=OXO.rstrip('/')):

tests/test_cli.py

@@ -15,7 +15,7 @@ def test_file_batch_input(rootdir):
     result = runner.invoke(ontoma, [test_file,'-'])
     assert result.exit_code == 0
     assert 'http://www.ebi.ac.uk/efo/EFO_0000270' in result.output
-    assert 'http://www.orpha.net/ORDO/Orphanet_309842' in result.output
+    assert 'http://purl.obolibrary.org/obo/MONDO_0002279' in result.output
 
 def test_batch_matching(rootdir):
 

tests/test_ontoma.py

@@ -10,9 +10,12 @@ def test_find_term_excludes(ontclient):
     assert not ontclient.find_term('breast')
 
 def test_suggest_hp_term_not_excluded(ontclient):
-    assert ontclient.find_term('hypogammaglobulinemia') == 'http://purl.obolibrary.org/obo/HP_0004313'
+    assert ontclient.find_term('hypogammaglobulinemia') == 'http://www.orpha.net/ORDO/Orphanet_229720'
 
 def test_catch_ordo(ontclient):
-    assert ontclient.find_term('Camptodactyly-arthropathy-coxa-vara-pericarditis syndrome') == 'http://www.orpha.net/ORDO/Orphanet_2848'
-    assert not ontclient.find_term('208250')
-    assert ontclient.find_term('208250',suggest=True) == 'http://www.orpha.net/ORDO/Orphanet_2848'
+    assert ontclient.find_term('Camptodactyly-arthropathy-coxa-vara-pericarditis syndrome') == 'http://www.ebi.ac.uk/efo/EFO_0009028'
+    assert ontclient.find_term('208250') == 'http://www.ebi.ac.uk/efo/EFO_0009028'
+    assert ontclient.find_term('208250',suggest=True) == 'http://www.ebi.ac.uk/efo/EFO_0009028'
+
+def test_query_comma(ontclient):
+    assert ontclient.find_term('3-methylglutaconic aciduria, type III') == 'http://www.orpha.net/ORDO/Orphanet_67047'

tests/test_phewascat.py

@@ -2,14 +2,13 @@ def test_find_term_asthma(ontclient):
     assert ontclient.find_term('asthma') == 'http://www.ebi.ac.uk/efo/EFO_0000270'
 
 def test_efo_direct_match(ontclient):
-    assert ontclient.find_term('Dementias') == 'http://purl.obolibrary.org/obo/HP_0000726'
+    assert ontclient.find_term('Dementias') == 'http://www.ebi.ac.uk/efo/EFO_0004718'
 
 def test_otzooma_mappings_whitespace(ontclient):
-    assert ontclient.find_term('Prostate cancer') == 'http://www.ebi.ac.uk/efo/EFO_0001663'
+    assert ontclient.find_term('Prostate cancer') == 'http://purl.obolibrary.org/obo/MONDO_0008315'
 
 def test_efo_match_with_apostrophe(ontclient):
     assert ontclient.find_term('Alzheimer\'s disease') == 'http://www.ebi.ac.uk/efo/EFO_0000249'
-    assert ontclient.find_term('290.1', code="ICD9CM") == 'http://www.ebi.ac.uk/efo/EFO_0000249'
 
 def test_match_with_hpo(ontclient):
     assert ontclient.find_term('Jaundice') == 'http://purl.obolibrary.org/obo/HP_0000952'