@@ -41,8 +41,8 @@ def _parse_ontology_name(self, term_id: str) -> str:
4141
4242 def get_term_ancestors (self , term_id : str , include_self : bool = False ) -> List [str ]:
4343 """
44- Get the ancestor ontology terms for a given term. If include_self is True, the term itself will be included as an
45- ancestor.
44+ Get the ancestor ontology terms for a given term. If include_self is True, the term itself will be included as
45+ an ancestor.
4646
4747 Example: get_term_ancestors("CL:0000005") -> ["CL:0000000", ...]
4848
@@ -51,13 +51,30 @@ def get_term_ancestors(self, term_id: str, include_self: bool = False) -> List[s
5151 :return: flattened List[str] of ancestor terms
5252 """
5353 ontology_name = self ._parse_ontology_name (term_id )
54- ancestors : List [ str ] = self .cxg_schema .ontology (ontology_name )[term_id ]["ancestors" ]
54+ ancestors = list ( self .cxg_schema .ontology (ontology_name )[term_id ]["ancestors" ]. keys ())
5555 return ancestors + [term_id ] if include_self else ancestors
5656
57- def get_term_list_ancestors (self , term_ids : str , include_self : bool = False ) -> Dict [str , List [ str ] ]:
57+ def get_term_ancestors_with_distances (self , term_id : str , include_self : bool = False ) -> Dict [str , int ]:
5858 """
59- Get the ancestor ontology terms for each term in a list. If include_self is True, the term itself will be included
60- as an ancestor.
59+ Get the ancestor ontology terms for a given term, and their distance from the term_id. If include_self is True,
60+ the term itself will be included as an ancestor.
61+
62+ Example: get_term_ancestors_with_distances("CL:0000005") -> {"CL:0000000": 1, ...}
63+
64+ :param term_id: str ontology term to find ancestors for
65+ :param include_self: boolean flag to include the term itself as an ancestor
66+ :return: Dict[str, int] map of ancestor terms and their respective distances from the term_id
67+ """
68+ ontology_name = self ._parse_ontology_name (term_id )
69+ ancestors : Dict [str , int ] = self .cxg_schema .ontology (ontology_name )[term_id ]["ancestors" ]
70+ if include_self :
71+ ancestors [term_id ] = 0
72+ return ancestors
73+
74+ def get_term_list_ancestors (self , term_ids : List [str ], include_self : bool = False ) -> Dict [str , List [str ]]:
75+ """
76+ Get the ancestor ontology terms for each term in a list. If include_self is True, the term itself will be
77+ included as an ancestor.
6178
6279 Example: get_term_list_ancestors(["CL:0000003", "CL:0000005"], include_self=True) -> {
6380 "CL:0000003": ["CL:0000003"],
@@ -71,10 +88,106 @@ def get_term_list_ancestors(self, term_ids: str, include_self: bool = False) ->
7188 """
7289 return {term_id : self .get_term_ancestors (term_id , include_self ) for term_id in term_ids }
7390
91+ def map_high_level_terms (self , term_ids : List [str ], high_level_terms : List [str ]) -> Dict [str , List [str ]]:
92+ """
93+ Given a list of ontology term IDs and a list of high_level_terms to map them to, returns a dictionary with
94+ format
95+
96+ {"CL:0000003": ["CL:0000000", ...], "CL:0000005": ["CL:0000000", ...]}
97+
98+ Where each term_id is mapped to a List[str] of high-level terms that it is a descendant of. Includes self
99+ as a descendant.
100+
101+ :param term_ids: list of str ontology terms to map high level terms for
102+ :param high_level_terms: list of str ontology terms that can be mapped to descendant term_ids
103+ :return: Dictionary mapping str term IDs to their respective List[str] of ancestor terms from the input list.
104+ Each key maps to empty list if there are no ancestors among the provided input.
105+ """
106+ ancestors = self .get_term_list_ancestors (term_ids , include_self = True )
107+ for term_id in term_ids :
108+ ancestors [term_id ] = [
109+ high_level_term for high_level_term in ancestors [term_id ] if high_level_term in high_level_terms
110+ ]
111+ return ancestors
112+
113+ def get_distance_between_terms (self , term_id_1 : str , term_id_2 : str ) -> int :
114+ """
115+ Get the distance between two ontology terms. The distance is defined as the number of edges between the
116+ two terms. Terms must be from the same ontology. Returns -1 if terms are disjoint.
117+
118+ :param term_id_1: str ontology term to find distance for
119+ :param term_id_2: str ontology term to find distance for
120+ :return: int distance between the two terms, measured in number of edges between their shortest path.
121+ """
122+ lcas = self .get_lowest_common_ancestors (term_id_1 , term_id_2 )
123+ if not lcas :
124+ return - 1
125+ return int (
126+ self .get_term_ancestors_with_distances (term_id_1 , include_self = True )[lcas [0 ]]
127+ + self .get_term_ancestors_with_distances (term_id_2 , include_self = True )[lcas [0 ]]
128+ )
129+
130+ def get_lowest_common_ancestors (self , term_id_1 : str , term_id_2 : str ) -> List [str ]:
131+ """
132+ Get the lowest common ancestors between two ontology terms that is from the given ontology.
133+ Terms must be from the same ontology. Ontologies are DAGs, so there may be multiple lowest common ancestors.
134+
135+ :param term_id_1: str ontology term to find LCA for
136+ :param term_id_2: str ontology term to find LCA for
137+ :return: str term ID of the lowest common ancestor term
138+ """
139+ # include path to term itself
140+ ontology = self ._parse_ontology_name (term_id_1 )
141+ if ontology != self ._parse_ontology_name (term_id_2 ):
142+ return []
143+ ancestors_1 = self .get_term_ancestors_with_distances (term_id_1 , include_self = True )
144+ ancestors_2 = self .get_term_ancestors_with_distances (term_id_2 , include_self = True )
145+ common_ancestors = set (ancestors_1 .keys ()) & set (ancestors_2 .keys ())
146+ min_sum_distances = float ("inf" )
147+ for ancestors in common_ancestors :
148+ sum_distances = ancestors_1 [ancestors ] + ancestors_2 [ancestors ]
149+ if sum_distances < min_sum_distances :
150+ min_sum_distances = sum_distances
151+ return [
152+ ancestor
153+ for ancestor in common_ancestors
154+ if ancestors_1 [ancestor ] + ancestors_2 [ancestor ] == min_sum_distances
155+ ]
156+
157+ def map_highest_level_term (self , term_ids : List [str ], high_level_terms : List [str ]) -> Dict [str , Union [str , None ]]:
158+ """
159+ Given a list of ontology term IDs and a list of high_level_terms to map them to, returns a dictionary with
160+ format
161+
162+ {"CL:0000003": "CL:0000000", "CL:0000005": "CL:0000000"}
163+
164+ Where each term_id is mapped to the highest level term that it is a descendant of, from the list provided. Includes
165+ term itself as a descendant. Maps to None if term_id does not map to any high level terms among the provided input.
166+
167+ :param term_ids: list of str ontology terms to map high level terms for
168+ :param high_level_terms: list of str ontology terms that can be mapped to descendant term_ids
169+ :return: Dictionary mapping str term IDs to their respective List[str] of ancestor terms from the input list.
170+ Each key maps to empty list if there are no ancestors among the provided input.
171+ """
172+ high_level_term_map = self .map_high_level_terms (term_ids , high_level_terms )
173+ highest_level_term_map = dict ()
174+ for term_id in term_ids :
175+ term_ancestors_and_distances = self .get_term_ancestors_with_distances (term_id , include_self = True )
176+ # map term_id to the high_level_term with the longest distance from term_id
177+ highest_level_term_map [term_id ] = (
178+ max (
179+ high_level_term_map [term_id ],
180+ key = lambda high_level_term : term_ancestors_and_distances [high_level_term ],
181+ )
182+ if high_level_term_map [term_id ]
183+ else None
184+ )
185+ return highest_level_term_map
186+
74187 def get_terms_descendants (self , term_ids : List [str ], include_self : bool = False ) -> Dict [str , List [str ]]:
75188 """
76- Get the descendant ontology terms for each term in a list. If include_self is True, the term itself will be included
77- as a descendant.
189+ Get the descendant ontology terms for each term in a list. If include_self is True, the term itself will be
190+ included as a descendant.
78191
79192 Example: get_terms_descendants(["CL:0000003", "CL:0000005"], include_self=True) -> {
80193 "CL:0000003": ["CL:0000003", "CL:0000004", ...],
@@ -83,8 +196,8 @@ def get_terms_descendants(self, term_ids: List[str], include_self: bool = False)
83196
84197 :param term_ids: list of str ontology terms to find descendants for
85198 :param include_self: boolean flag to include the term itself as an descendant
86- :return: Dictionary mapping str term IDs to their respective flattened List[str] of descendant terms. Maps to empty
87- list if there are no descendants.
199+ :return: Dictionary mapping str term IDs to their respective flattened List[str] of descendant terms. Maps to
200+ empty list if there are no descendants.
88201 """
89202 descendants_dict = dict ()
90203 ontology_names = set ()
@@ -96,7 +209,8 @@ def get_terms_descendants(self, term_ids: List[str], include_self: bool = False)
96209 for ontology in ontology_names :
97210 for candidate_descendant , candidate_metadata in self .cxg_schema .ontology (ontology ).items ():
98211 for ancestor_id in descendants_dict :
99- if ancestor_id in candidate_metadata ["ancestors" ]:
212+ ancestors = candidate_metadata ["ancestors" ].keys ()
213+ if ancestor_id in ancestors :
100214 descendants_dict [ancestor_id ].append (candidate_descendant )
101215
102216 return descendants_dict
0 commit comments