Merge pull request #185 from monarch-initiative/variant_formatter

Created a Formatter

src/genophenocorr/model/__init__.py

@@ -16,7 +16,7 @@
 __all__ = [
     'Cohort', 'Patient', 'SampleLabels',
     'Phenotype', 'Disease',
-    'Variant', 'VariantCoordinates', 'Genotype', 'Genotypes', 'Genotyped',
+    'Variant', 'VariantCoordinates', 'Genotype', 'Genotypes', 'Genotyped', 
     'TranscriptAnnotation', 'VariantEffect', 'TranscriptInfoAware', 'TranscriptCoordinates',
     'ProteinMetadata', 'ProteinFeature', 'FeatureInfo', 'FeatureType',
 ]

src/genophenocorr/model/_cohort.py

@@ -1,7 +1,7 @@
 import itertools
 import typing
 
-from collections import Counter, defaultdict
+from collections import Counter
 
 import hpotk
 

src/genophenocorr/model/_variant.py

@@ -6,7 +6,6 @@
 
 from .genome import Region, GenomicRegion
 from ._gt import Genotyped, Genotypes
-from ._protein import ProteinMetadata
 from ._variant_effects import VariantEffect
 
 
@@ -262,7 +261,17 @@ def variant_key(self) -> str:
         if self.is_structural():
             return f'{self.chrom}_{self.start + 1}_{self.end}_{self.alt[1:-1]}'
         else:
-            return f'{self.chrom}_{self.start + 1}_{self.end}_{self.ref}_{self.alt}'
+            key = f'{self.chrom}_{self.start + 1}_{self.end}_{self.ref}_{self.alt}'
+            if len(key) > 50:
+                ref = None
+                alt = None
+                if len(self.ref) > 10:
+                    ref = f"--{len(self.ref)}bp--"
+                if len(self.alt) > 10:
+                    alt = f"--{len(self.alt)}bp--"
+                return f"{self.chrom}_{self.start + 1}_{self.end}_{ref if not None else self.ref}_{alt if not None else self.alt}"
+            else:
+                return key
 
     @property
     def variant_class(self) -> str:
@@ -385,6 +394,22 @@ def get_hgvs_cdna_by_tx_id(self, transcript_id:str) -> typing.Optional[str]:
             if tx_ann.transcript_id == transcript_id:
                 return tx_ann.hgvs_cdna
         return None
+
+    def get_preferred_tx_annotation(self) -> typing.Optional[TranscriptAnnotation]:
+        """Get the `TranscriptAnnotation` that represents the result of the functional annotation
+        with respect to the preferred transcript of a gene. 
+        
+        Returns `None` if transcript annotations is no preferred transcript found. 
+
+        Returns:
+            typing.Optional[TranscriptAnnotation]: The `TranscriptAnnotation` with respect 
+                                                   to the preferred transcript 
+                                                   or `None` if the preferred transcript info is not available.
+        """
+        for tx in self.tx_annotations:
+            if tx.is_preferred:
+                return tx
+        return None
 
 
 class Variant(VariantCoordinateAware, FunctionalAnnotationAware, Genotyped):
@@ -470,3 +495,4 @@ def __str__(self) -> str:
         return (f"Variant(variant_coordinates:{str(self.variant_coordinates)}, "
                 f"tx_annotations:{self.tx_annotations}, "
                 f"genotypes:{self.genotypes})")
+

src/genophenocorr/view/__init__.py

@@ -5,11 +5,13 @@
 from ._stats import StatsViewer
 from ._txp import VariantTranscriptVisualizer
 from ._protein_visualizer import ProteinVisualizer
+from ._formatter import VariantFormatter
 
 __all__ = [
     'CohortViewable',
     'ProteinVisualizer', 'ProteinVisualizable', 'ProteinViewable',
     'DiseaseViewable',
     'StatsViewer',
-    'VariantTranscriptVisualizer'
+    'VariantTranscriptVisualizer',
+    'VariantFormatter',
 ]

src/genophenocorr/view/_cohort.py

@@ -3,7 +3,8 @@
 from hpotk import MinimalOntology
 from jinja2 import Environment, PackageLoader
 
-from genophenocorr.model import Cohort, Variant
+from genophenocorr.model import Cohort
+from ._formatter import VariantFormatter
 
 
 class CohortViewable:
@@ -33,7 +34,7 @@ def __init__(
     def process(
             self,
             cohort: Cohort,
-            transcript_id: str
+            transcript_id: typing.Optional[str] = None
     ) -> str:
         """
         Create an HTML that should be shown with display(HTML(..)) of the ipython package.
@@ -51,7 +52,7 @@ def process(
     def _prepare_context(
             self,
             cohort: Cohort,
-            transcript_id: str
+            transcript_id: typing.Optional[str]
     ) -> typing.Mapping[str, typing.Any]:
 
         hpo_counts = list()
@@ -95,6 +96,8 @@ def _prepare_context(
                         var_effects_list.append({"effect": effect, "count": count})
         else:
             has_transcript = False
+        if transcript_id is None:
+            transcript_id = "MANE transcript ID"
         # The following dictionary is used by the Jinja2 HTML template
         return {
             "n_individuals": len(cohort.all_patients),
@@ -112,17 +115,10 @@ def _prepare_context(
             "transcript_id": transcript_id,
         }
 
-    @staticmethod
-    def get_display(variant: Variant, transcript_id: str) -> str:
-        for annot in variant.tx_annotations:
-            if annot.transcript_id == transcript_id:
-                return annot.hgvs_cdna
-        return variant.variant_coordinates.variant_key
-
     @staticmethod
     def get_variant_description(
             cohort: Cohort,
-            transcript_id: str,
+            transcript_id: typing.Optional[str],
             only_hgvs: bool = True,
     ) -> typing.Mapping[str, str]:
         """
@@ -137,9 +133,10 @@ def get_variant_description(
         """
         chrom_to_display = dict()
         all_var_set = cohort.all_variants()
+        var_formatter = VariantFormatter(transcript_id)
         for var in all_var_set:
             var_string = var.variant_coordinates.variant_key
-            display = CohortViewable.get_display(variant=var, transcript_id=transcript_id)
+            display = var_formatter.format_as_string(var)
             if only_hgvs:
                 # do not show the transcript id
                 fields = display.split(":")

src/genophenocorr/view/_formatter.py

@@ -0,0 +1,50 @@
+import abc
+import typing
+
+from genophenocorr.model import Variant
+
+T = typing.TypeVar('T')
+
+class Formatter(typing.Generic[T], metaclass=abc.ABCMeta):
+
+    @abc.abstractmethod
+    def format_as_string(self, item: T) -> str:
+        """ Inputs an item and outputs a human readable string that can be used
+        to more easily read the item in tables or other visualizers.
+
+        Args:
+            item (T): an element to be formatted
+
+        Returns:
+            str: a human readable string 
+        """
+        pass
+
+class VariantFormatter(Formatter[Variant]):
+    """
+    A class that can be used to format a `Variant` to a human readable string
+    """
+    def __init__(self, tx_id: typing.Optional[str] = None) -> None:
+        self._tx_id = tx_id
+
+    def format_as_string(self, item: Variant) -> str:
+        """ 
+        Args:
+            item (Variant): An object of class `Variant` representing a variant.
+
+        Returns:
+            str: A human readable string for the variant.
+        """
+        if self._tx_id is not None:
+            transcript = item.get_tx_anno_by_tx_id(self._tx_id)
+        else:
+            transcript = item.get_preferred_tx_annotation()
+        if transcript is not None and transcript.hgvs_cdna is not None:
+            if len(transcript.hgvs_cdna) > 50:
+                return "Long HGVS"
+            return transcript.hgvs_cdna
+        elif item.variant_coordinates.variant_key is not None:
+            return item.variant_coordinates.variant_key
+        else:
+            # To be reevaluated
+            return f"Variant {item} has no string format."

tests/model/test_variant.py

@@ -4,7 +4,6 @@
 
 from genophenocorr.model import Variant, Cohort
 
-
 class TestVariant:
 
     @pytest.fixture
@@ -28,4 +27,4 @@ def test_get_hgvs_cdna_by_tx(
     ):
         hgvs = some_variant.get_hgvs_cdna_by_tx_id(transcript_id=tx_id)
 
-        assert hgvs == expected
+        assert hgvs == expected

tests/view/test_formatter.py

@@ -0,0 +1,24 @@
+import pytest
+
+import typing
+
+from genophenocorr.model import Cohort
+from genophenocorr.view import VariantFormatter
+
+class TestFormatter:
+
+    @pytest.mark.parametrize(
+        "variant, expected",
+        [
+            ('12_56004525_56004525_A_G', "NM_001032386.2:c.1136A>G"),
+        ]
+    )
+    def test_variant_formatter(
+        self, 
+        variant: str,
+        expected: str,
+        suox_cohort: Cohort
+    ):
+        var = suox_cohort.get_variant_by_key(variant)
+        formatter = VariantFormatter()
+        assert formatter.format_as_string(var) == expected