age output

src/pyphetools/__init__.py

@@ -3,7 +3,7 @@
 from . import visualization
 from . import validation
 
-__version__ = "0.9.50"
+__version__ = "0.9.51"
 
 __all__ = [
     "creation",

src/pyphetools/creation/pyphetools_age.py

@@ -68,6 +68,18 @@ def is_valid(self) -> bool:
         """
         pass
 
+    @abc.abstractmethod
+    def to_hpo_age(self) -> "HpoAge":
+        """convert this PyPheToolsAge object to an HpoAge object.
+        This is the identity function if the object already is of class HpoAge. For other classes, we choose the
+        HpoAge object that is most appropriate given the ISO8601 or Gestational age. The purpose of this
+        function is that for HPOA output, we need to have an HPO term to denote the age of onset frequencies
+
+        :returns: HpoAge object representing the age
+        :rtype: HpoAge
+        """
+        pass
+
     @property
     def age_string(self):
         if self.is_valid():
@@ -115,6 +127,12 @@ def to_ga4gh_time_element(self):
     def is_valid(self):
         return False
 
+    def to_hpo_age(self):
+        """There is no information about age, so return the NoneAge object to denote this.
+        Client code should always check the is_valid function
+        """
+        return self
+
 
 class IsoAge(PyPheToolsAge):
     """Class to record and sort ages formated according to iso8601
@@ -188,6 +206,29 @@ def to_iso8601(self):
         else:
             return "".join(components)
 
+    def to_hpo_age(self):
+        """Convert to HpoAge object
+        """
+        if self._years >= 60:
+            return HpoAge("Late onset")
+        elif self._years >= 40:
+            return HpoAge("Middle age onset")
+        elif self._years >= 16:
+            return HpoAge("Young adult onset")
+        elif self._years >= 5:
+            return HpoAge("Juvenile onset")
+        elif self._years >= 1:
+            return HpoAge("Childhood onset")
+        elif self._months >= 1:
+            return HpoAge("Infantile onset")
+        elif self._days >= 1:
+            return HpoAge("Neonatal onset")
+        elif self._days == 0:
+            return HpoAge("Congenital onset")
+        else:
+            raise ValueError(f"[ERROR] Could not calculate HpoAge for {self.age_string}")
+
+
     def to_ga4gh_time_element(self) -> PPKt.TimeElement:
         """
         :returns: a representation of Age formated as one of the options of GA4GH TimeElement
@@ -255,6 +296,11 @@ def to_ga4gh_time_element(self) -> PPKt.TimeElement:
         time_elem.ontology_class.CopyFrom(clz)
         return time_elem
 
+    def to_hpo_age(self):
+        """Return self, this is already an HpoAge object
+        """
+        return self
+
     def is_valid(self):
         return True
 
@@ -284,6 +330,21 @@ def to_ga4gh_time_element(self) -> PPKt.TimeElement:
         time_elem.age.gestational_age.CopyFrom(gest_age)
         return time_elem
 
+    def to_hpo_age(self):
+        """Return self, this is already an HpoAge object
+        """
+        if self._weeks >= 28:
+            # prior to birth during the third trimester, which is defined as 28 weeks and zero days (28+0) of gestation and beyond.
+            return HpoAge("Third trimester onset") # HP:0034197
+        elif self._weeks >= 14:
+            # prior to birth during the second trimester, which comprises the range of gestational ages from 14 0/7 weeks to 27 6/7 (inclusive).
+            return HpoAge("Second trimester onset") # HP:0034198
+        elif self._weeks >= 11:
+            # 11 0/7 to 13 6/7 weeks of gestation (inclusive).
+            return HpoAge("Late first trimester onset")  #  HP:0034199
+        else:
+            return HpoAge("Embryonal onset")
+
     @staticmethod
     def is_gestational_age(age_string):
         """Gestational age should be formated as W+D, e.g. 33+2

src/pyphetools/visualization/onset_calculator.py

@@ -1,10 +1,12 @@
 from collections import defaultdict
-from typing import Dict, List, Optional, Union
-from ..creation.constants import Constants
-from ..creation.pyphetools_age import PyPheToolsAge
+from typing import Dict, List
+import re
+from ..creation.pyphetools_age import HPO_ONSET_TERMS
 from ..creation.hp_term import HpTerm
 
 
+
+
 class OnsetCalculator:
     """
     This class calculates the age of onset anntoations for the HPOA file
@@ -27,20 +29,80 @@ def __init__(self, phenopacket_list):
                 print("Warning: Identified multiple disease element")
             disease = ppack.diseases[0]
             if disease.HasField("onset"):
+                # onset is a GA4GH TimeElement
+                # In pyphetools, it can be an OntologyClass, an Age, or a GestationalAge
                 onset = disease.onset
                 if onset.HasField("ontology_class"):
-                    onset_term = onset.ontology_class
-                    hpo_onset_term = HpTerm(hpo_id=onset_term.id, label=onset_term.label)
-                    self._pmid_to_onsetlist_d[pmid].append(hpo_onset_term)
-                elif  onset.HasField("age"):
-                    isoage = onset.age.iso8601duration
-                    hpo_onset_term = PyPheToolsAge.onset_to_hpo_term(onset_string=isoage)
-                    if hpo_onset_term is not None and hpo_onset_term != Constants.NOT_PROVIDED:
+                        onset_term = onset.ontology_class
+                        hpo_onset_term = HpTerm(hpo_id=onset_term.id, label=onset_term.label)
                         self._pmid_to_onsetlist_d[pmid].append(hpo_onset_term)
+                elif onset.HasField("age"):
+                    hpo_onset_term = self._get_hpo_onset_term_from_iso8601(onset.age.iso8601duration)
+                    self._pmid_to_onsetlist_d[pmid].append(hpo_onset_term)
+                elif onset.HasField("gestational_age"):
+                    hpo_onset_term = self._get_hpo_onset_term_from_gestational_age(onset.age.iso8601duration)
+                    self._pmid_to_onsetlist_d[pmid].append(hpo_onset_term)
                 else:
-                    print("[ERROR] Could not parse disease onset")
-                    print(disease.onset)
-                    raise ValueError(f"Could not parse disease onset {disease}")
+                    raise ValueError(f"onset was present but could not be decoded: {onset}")
+
+
+    def _get_hpo_onset_term_from_iso8601(self, isostring):
+        # the following regex gets years, months, days - optionally (when we get to this point in pyphetools, we cannot have weeks)
+        ISO8601_REGEX = r"^P(\d+Y)?(\d+M)?(\d+D)?"
+        match = re.search(ISO8601_REGEX, isostring)
+        if match:
+            y = match.group(1) or "0Y"
+            m = match.group(2) or "0M"
+            d = match.group(3) or "0D"
+            y = int(y[0:-1]) # all but last character
+            m = int(m[0:-1])
+            d = int(d[0:-1])
+            label = None
+            if y >= 60:
+                label = "Late onset"
+            elif y >= 40:
+                label = "Middle age onset"
+            elif y >= 16:
+                label = "Young adult onset"
+            elif y >= 5:
+                label = "Juvenile onset"
+            elif y >= 1:
+                label = "Childhood onset"
+            elif m >= 1:
+                label = "Infantile onset"
+            elif d  >= 1:
+                label = "Neonatal onset"
+            elif d == 0:
+                label = "Congenital onset"
+            else:
+                raise ValueError(f"[ERROR] Could not parse iso8601 \"{isostring}\"")
+            if label not in HPO_ONSET_TERMS:
+                # should never happen ...
+                raise ValueError(f"Could not identify onset label {label}")
+            hpo_id = HPO_ONSET_TERMS.get(label)
+            return HpTerm(hpo_id=hpo_id, label=label)
+
+    def _get_hpo_onset_term_from_gestational_age(self, gestational_age):
+        weeks = gestational_age.weeks
+        # days not relevant to identifying the HPO Onset term
+        label = None
+        if weeks >= 28:
+            # prior to birth during the third trimester, which is defined as 28 weeks and zero days (28+0) of gestation and beyond.
+            label = "Third trimester onset" # HP:0034197
+        elif weeks >= 14:
+            # prior to birth during the second trimester, which comprises the range of gestational ages from 14 0/7 weeks to 27 6/7 (inclusive).
+            label = "Second trimester onset" # HP:0034198
+        elif weeks >= 11:
+            # 11 0/7 to 13 6/7 weeks of gestation (inclusive).
+            label = "Late first trimester onset"  #  HP:0034199
+        else:
+            label = "Embryonal onset"
+        if label not in HPO_ONSET_TERMS:
+                # should never happen ...
+                raise ValueError(f"Could not identify onset label {label}")
+        hpo_id = HPO_ONSET_TERMS.get(label)
+        return HpTerm(hpo_id=hpo_id, label=label)
+
 
     def get_pmid_to_onset_d(self)-> Dict[str, List[HpTerm]]:
         return self._pmid_to_onsetlist_d