Refactor Anopheles tests on genome sequence, genome features and samp…

…le metadata (#399)

* remove dup tests

* refactor fixtures

* fix tests

* remove dup tests

* update poetry

* remove dup tests

* sample metadata test refactoring

* refactor snp_data tests

* refactor sample metadata tests

tests/anoph/conftest.py

@@ -3,6 +3,7 @@
 import string
 from pathlib import Path
 from random import choice, choices, randint
+from typing import Any, Dict, Tuple
 
 import numpy as np
 import pandas as pd
@@ -495,15 +496,25 @@ def simulate_site_annotations(path, genome):
     zarr.consolidate_metadata(path)
 
 
-class Ag3Simulator:
-    def __init__(self, fixture_dir):
+class AnophelesSimulator:
+    def __init__(
+        self,
+        fixture_dir: Path,
+        bucket: str,
+        releases: Tuple[str, ...],
+        has_aims: bool,
+        has_cohorts_by_quarter: bool,
+    ):
         self.fixture_dir = fixture_dir
-        self.bucket = "vo_agam_release"
+        self.bucket = bucket
         self.bucket_path = (self.fixture_dir / "simulated" / self.bucket).resolve()
         self.results_cache_path = (
-            self.fixture_dir / "simulated" / "ag3_results_cache"
+            self.fixture_dir / "simulated" / "results_cache"
         ).resolve()
         self.url = self.bucket_path.as_uri()
+        self.releases = releases
+        self.has_aims = has_aims
+        self.has_cohorts_by_quarter = has_cohorts_by_quarter
 
         # Clear out the fixture directories.
         shutil.rmtree(self.bucket_path, ignore_errors=True)
@@ -512,9 +523,14 @@ def __init__(self, fixture_dir):
         # Ensure the fixture directory exists.
         self.bucket_path.mkdir(parents=True, exist_ok=True)
 
+        # These members to be overridden/populated in subclasses.
+        self.config: Dict[str, Any] = dict()
+        self.contig_sizes: Dict[str, int] = dict()
+        self.release_manifests: Dict[str, pd.DataFrame] = dict()
+        self.genome = None
+        self.genome_features = None
+
         # Create fixture data.
-        self.releases = ("3.0", "3.1")
-        self.release_manifests = dict()
         self.init_config()
         self.init_public_release_manifest()
         self.init_pre_release_manifest()
@@ -526,6 +542,62 @@ def __init__(self, fixture_dir):
         self.init_snp_genotypes()
         self.init_site_annotations()
 
+    @property
+    def contigs(self) -> Tuple[str, ...]:
+        return tuple(self.config["CONTIGS"])
+
+    def random_contig(self):
+        return choice(self.contigs)
+
+    def random_region_str(self):
+        contig = self.random_contig()
+        contig_size = self.contig_sizes[contig]
+        region_start = randint(1, contig_size)
+        region_end = randint(region_start, contig_size)
+        region = f"{contig}:{region_start:,}-{region_end:,}"
+        return region
+
+    def init_config(self):
+        pass
+
+    def init_public_release_manifest(self):
+        pass
+
+    def init_pre_release_manifest(self):
+        pass
+
+    def init_genome_sequence(self):
+        pass
+
+    def init_genome_features(self):
+        pass
+
+    def init_metadata(self):
+        pass
+
+    def init_snp_sites(self):
+        pass
+
+    def init_site_filters(self):
+        pass
+
+    def init_snp_genotypes(self):
+        pass
+
+    def init_site_annotations(self):
+        pass
+
+
+class Ag3Simulator(AnophelesSimulator):
+    def __init__(self, fixture_dir):
+        super().__init__(
+            fixture_dir=fixture_dir,
+            bucket="vo_agam_release",
+            releases=("3.0", "3.1"),
+            has_aims=True,
+            has_cohorts_by_quarter=True,
+        )
+
     def init_config(self):
         self.config = {
             "PUBLIC_RELEASES": ["3.0"],
@@ -547,10 +619,6 @@ def init_config(self):
         with config_path.open(mode="w") as f:
             json.dump(self.config, f, indent=4)
 
-    @property
-    def contigs(self):
-        return tuple(self.config["CONTIGS"])
-
     def init_public_release_manifest(self):
         # Here we create a release manifest for an Ag3-style
         # public release. Note this is not the exact same data
@@ -821,35 +889,15 @@ def init_site_annotations(self):
         simulate_site_annotations(path=path, genome=self.genome)
 
 
-class Af1Simulator:
+class Af1Simulator(AnophelesSimulator):
     def __init__(self, fixture_dir):
-        self.fixture_dir = fixture_dir
-        self.bucket = "vo_afun_release"
-        self.bucket_path = (self.fixture_dir / "simulated" / self.bucket).resolve()
-        self.url = self.bucket_path.as_uri()
-        self.results_cache_path = (
-            self.fixture_dir / "simulated" / "af1_results_cache"
-        ).resolve()
-
-        # Clear out the fixture directories.
-        shutil.rmtree(self.bucket_path, ignore_errors=True)
-        shutil.rmtree(self.results_cache_path, ignore_errors=True)
-
-        # Ensure the fixture directory exists.
-        self.bucket_path.mkdir(parents=True, exist_ok=True)
-
-        # Create fixture data.
-        self.releases = ("1.0",)
-        self.release_manifests = dict()
-        self.init_config()
-        self.init_public_release_manifest()
-        self.init_genome_sequence()
-        self.init_genome_features()
-        self.init_metadata()
-        self.init_snp_sites()
-        self.init_site_filters()
-        self.init_snp_genotypes()
-        self.init_site_annotations()
+        super().__init__(
+            fixture_dir=fixture_dir,
+            bucket="vo_afun_release",
+            releases=("1.0",),
+            has_aims=False,
+            has_cohorts_by_quarter=False,
+        )
 
     def init_config(self):
         self.config = {
@@ -871,10 +919,6 @@ def init_config(self):
         with config_path.open(mode="w") as f:
             json.dump(self.config, f, indent=4)
 
-    @property
-    def contigs(self):
-        return tuple(self.config["CONTIGS"])
-
     def init_public_release_manifest(self):
         # Here we create a release manifest for an Af1-style
         # public release. Note this is not the exact same data

tests/anoph/test_genome_features.py

@@ -8,6 +8,7 @@
 from malariagen_data import af1 as _af1
 from malariagen_data import ag3 as _ag3
 from malariagen_data.anoph.genome_features import AnophelesGenomeFeaturesData
+from malariagen_data.util import Region, resolve_region
 
 
 @pytest.fixture
@@ -99,19 +100,29 @@ def test_genome_features_region_contig(fixture, api: AnophelesGenomeFeaturesData
 
 @parametrize_with_cases("fixture,api", cases=".")
 def test_genome_features_region_string(fixture, api: AnophelesGenomeFeaturesData):
-    for contig in fixture.contigs:
-        contig_size = fixture.contig_sizes[contig]
-        region_start = random.randint(1, contig_size)
-        region_end = random.randint(region_start, contig_size)
-        region = f"{contig}:{region_start:,}-{region_end:,}"
+    parametrize_region = [
+        # Single contig.
+        fixture.random_contig(),
+        # List of contigs.
+        [fixture.random_contig(), fixture.random_contig()],
+        # Single region.
+        fixture.random_region_str(),
+        # List of regions.
+        [fixture.random_region_str(), fixture.random_region_str()],
+    ]
+
+    for region in parametrize_region:
         df_gf = api.genome_features(region=region, attributes=None)
         expected_cols = gff3_cols + ["attributes"]
         assert df_gf.columns.to_list() == expected_cols
         # N.B., it's possible that the region overlaps no features.
-        if len(df_gf) > 0:
-            assert (df_gf["contig"] == contig).all()
-            assert (df_gf["end"] >= region_start).all()
-            assert (df_gf["start"] <= region_end).all()
+        r = resolve_region(api, region)
+        if len(df_gf) > 0 and isinstance(r, Region):
+            assert (df_gf["contig"] == r.contig).all()
+            if r.start is not None:
+                assert (df_gf["end"] >= r.start).all()
+            if r.end is not None:
+                assert (df_gf["start"] <= r.end).all()
 
 
 @parametrize_with_cases("fixture,api", cases=".")