cleanup

src/pyphetools/__init__.py

@@ -5,7 +5,7 @@
 from . import validation
 
 
-__version__ = "0.9.115"
+__version__ = "0.9.116"
 
 
 __all__ = [

src/pyphetools/creation/__init__.py

@@ -23,12 +23,10 @@
 from .metadata import MetaData
 from .mode_of_inheritance import Moi
 from .ontology_terms import OntologyTerms
-from .option_column_mapper import OptionColumnMapper
 from .intergenic_variant import IntergenicVariant
 from .pyphetools_age import PyPheToolsAge, AgeSorter, HPO_ONSET_TERMS
 from .sex_column_mapper import SexColumnMapper
 from .simple_column_mapper import SimpleColumnMapper
-from .scm_generator import SimpleColumnMapperGenerator
 from .structural_variant import StructuralVariant
 from .thresholded_column_mapper import ThresholdedColumnMapper
 from .thresholder import Thresholder

src/pyphetools/creation/hpo_base_cr.py

@@ -3,10 +3,8 @@
 import abc
 from collections import defaultdict
 
-from .column_mapper import ColumnMapper
 from .hp_term import HpTerm
 from .hpo_cr import HpoConceptRecognizer
-from .simple_column_mapper import SimpleColumnMapper
 
 
 class ConceptMatch:
@@ -165,7 +163,7 @@ def parse_cell_for_exact_matches(self, cell_text, custom_d) -> typing.List[HpTer
             results.extend(self._get_non_overlapping_matches(hits=hits))
         return
 
-    def _get_non_overlapping_matches(self, hits: [typing.List[ConceptMatch]]) -> typing.List[HpTerm]:
+    def _get_non_overlapping_matches(self, hits: typing.List[ConceptMatch]) -> typing.List[HpTerm]:
         """The prupose of this method is to choose a list of non-overlapping matches
 
         Sometimes, we get multiple matches that partially overlap. We will greedily take the longest matches and discard overlaps.
@@ -225,21 +223,3 @@ def contains_term_label(self, hpo_label) -> bool:
         """
         return hpo_label.lower() in self._label_to_id
 
-    def initialize_simple_column_maps(self, column_name_to_hpo_label_map, observed, excluded):
-        if observed is None or excluded is None:
-            raise ValueError("Symbols for observed (e.g., +, Y, yes) and excluded (e.g., -, N, no) required")
-        if not isinstance(column_name_to_hpo_label_map, dict):
-            raise ValueError("column_name_to_hpo_label_map must be a dict with column to HPO label mappings")
-        simple_mapper_d = defaultdict(ColumnMapper)
-        for column_name, hpo_label_and_id in column_name_to_hpo_label_map.items():
-            if not isinstance(hpo_label_and_id, list):
-                raise ValueError(f"Expected {hpo_label_and_id} to be a two-item list with HPO label and id")
-            hpo_label = hpo_label_and_id[0]
-            expected_id = hpo_label_and_id[1]
-            hp_term = self.get_term_from_label(hpo_label)
-            if hp_term.id != expected_id:
-                raise ValueError(f"Got {hp_term.id} but was expecting {expected_id} for {hpo_label}")
-            mpr = SimpleColumnMapper(column_name=column_name, hpo_id=hp_term.id, hpo_label=hp_term.label,
-                                     observed=observed, excluded=excluded)
-            simple_mapper_d[column_name] = mpr
-        return simple_mapper_d

src/pyphetools/creation/hpo_cr.py

@@ -53,10 +53,3 @@ def get_term_from_label(self, label) -> HpTerm:
         :rtype: HpTerm
         """
         pass
-
-    @abc.abstractmethod
-    def initialize_simple_column_maps(self, column_name_to_hpo_label_map, observed, excluded, non_measured=None):
-        """
-        Create a dictionary of SimpleColumnMappers
-        """
-        pass

src/pyphetools/creation/pyphetools_age.py

@@ -7,7 +7,7 @@
 DAYS_IN_WEEK = 7
 AVERAGE_DAYS_IN_MONTH = 30.437
 AVERAGE_DAYS_IN_YEAR = 365.25
-
+ISO8601_REGEX = r"^P(\d+Y)?(\d+M)?(\d+D)?"
 
 from ..pp.v202 import OntologyClass as OntologyClass202
 from ..pp.v202 import TimeElement as TimeElement202
@@ -255,7 +255,10 @@ def get_age_pp201(age_string: str) -> typing.Optional[TimeElement202]:
         if isinstance(age_string, float) and math.isnan(age_string):
             return None  # sometimes pandas returns an empty cell as a float NaN
         if age_string.startswith("P"):
-            return TimeElement202(Age202(age_string))
+            if re.fullmatch(ISO8601_REGEX, age_string):
+                return TimeElement202(Age202(age_string))
+            else:
+                raise ValueError(f"Invalid ISO8601 expression: '{age_string}'")
         elif age_string in HPO_ONSET_TERMS:
             hpo_id = HPO_ONSET_TERMS.get(age_string)
             onsetClz = OntologyClass202(id=hpo_id, label=age_string)