Merge pull request #337 from monarch-initiative/investigate-qc

Fix Q/C reporting

pyproject.toml

@@ -33,6 +33,7 @@ classifiers = [
 ]
 dependencies = [
     "hpo-toolkit>=0.3.0",
+    "stairval>=0.2.0",
     "Jinja2>=3.1.4,<4.0.0",
     "protobuf>=3.15.0",
     "pandas>=2.0.0,<3.0.0",

src/gpsea/model/_cohort.py

@@ -267,12 +267,16 @@ def __hash__(self) -> int:
 
 
 class Cohort(typing.Sized, typing.Iterable[Patient]):
+    """
+    Cohort is a collection of individuals that have been preprocessed
+    and are ready for genotype-phenotype association analysis.
+    """
 
     @staticmethod
     def from_patients(
-            members: typing.Iterable[Patient],
-            include_patients_with_no_HPO: bool = False,
-            include_patients_with_no_variants: bool = False,
+        members: typing.Iterable[Patient],
+        include_patients_with_no_HPO: bool = False,
+        include_patients_with_no_variants: bool = False,
     ):
         """
         Create a cohort from a sequence of patients.

src/gpsea/preprocessing/__init__.py

@@ -1,7 +1,6 @@
 from ._api import PreprocessingValidationResult
 from ._api import TranscriptCoordinateService, GeneCoordinateService
 from ._api import VariantCoordinateFinder, FunctionalAnnotator, ImpreciseSvFunctionalAnnotator, ProteinMetadataService
-from ._audit import Auditor, DataSanityIssue, Level, Notepad, NotepadTree
 from ._config import load_phenopacket_folder, load_phenopacket_files, load_phenopackets
 from ._config import configure_caching_cohort_creator, configure_cohort_creator
 from ._config import configure_default_protein_metadata_service, configure_protein_metadata_service
@@ -27,6 +26,5 @@
     'UniprotProteinMetadataService',
     'VepFunctionalAnnotator', 'VariantAnnotationCache', 'VarCachingFunctionalAnnotator',
     'VVHgvsVariantCoordinateFinder', 'VVMultiCoordinateService',
-    'Auditor', 'DataSanityIssue', 'Level', 'Notepad', 'NotepadTree',
     'DefaultImpreciseSvFunctionalAnnotator',
 ]

src/gpsea/preprocessing/_api.py

@@ -1,9 +1,10 @@
 import abc
-import io
 import os
 import sys
 import typing
 
+from stairval.notepad import Notepad
+
 from gpsea.model import (
     VariantCoordinates,
     ProteinMetadata,
@@ -13,8 +14,6 @@
     ImpreciseSvInfo,
 )
 
-from ._audit import NotepadTree
-
 T = typing.TypeVar("T")
 
 
@@ -142,7 +141,7 @@ class PreprocessingValidationResult:
     def __init__(
         self,
         policy: str,
-        notepad: NotepadTree,
+        notepad: Notepad,
     ):
         self._policy = policy
         self._notepad = notepad
@@ -175,7 +174,7 @@ def is_ok(self) -> bool:
 
     def summarize(
         self,
-        file: io.TextIOBase = sys.stdout,
+        file: typing.TextIO = sys.stdout,
         indent: int = 2,
     ):
         """
@@ -189,13 +188,13 @@ def summarize(
         file.write(f"Validated under {self._policy} policy")
         file.write(os.linesep)
 
-        n_errors = sum(node.error_count() for node in self._notepad.iterate_nodes())
-        n_warnings = sum(node.warning_count() for node in self._notepad.iterate_nodes())
+        n_errors = sum(node.error_count() for node in self._notepad.iter_sections())
+        n_warnings = sum(node.warning_count() for node in self._notepad.iter_sections())
         if n_errors > 0 or n_warnings > 0:
             file.write("Showing errors and warnings")
             file.write(os.linesep)
 
-            for node in self._notepad.iterate_nodes():
+            for node in self._notepad.iter_sections():
                 if node.has_errors_or_warnings(include_subsections=True):
                     # We must report the node label even if there are no issues with the node.
                     l_pad = " " * (node.level * indent)
@@ -207,19 +206,19 @@ def summarize(
                         file.write(os.linesep)
                         for error in node.errors():
                             file.write(
-                                l_pad + " " + error.message + f". {error.solution}"
-                                if error.solution
-                                else ""
+                                l_pad + " ·" + error.message + (
+                                    f". {error.solution}" if error.solution else ""
+                                )
                             )
                             file.write(os.linesep)
                     if node.has_warnings():
                         file.write(l_pad + " warnings:")
                         file.write(os.linesep)
                         for warning in node.warnings():
                             file.write(
-                                l_pad + " ·" + warning.message + f". {warning.solution}"
-                                if warning.solution
-                                else ""
+                                l_pad + " ·" + warning.message + (
+                                    f". {warning.solution}" if warning.solution else ""
+                                )
                             )
                             file.write(os.linesep)
         else: