Update tests

src/stripepy/utils/finders.py

@@ -157,12 +157,12 @@ def find_HIoIs(
     if logger is None:
         logger = structlog.get_logger()
 
-    iterable_input = (
+    params = (
         (seed_site, seed_site_bounds[num_MP], seed_site_bounds[num_MP + 1])
         for num_MP, seed_site in enumerate(seed_sites)
     )
 
-    tasks = map_(partial(find_horizontal_domain, pseudodistribution, max_width=max_width), iterable_input)
+    tasks = map_(partial(find_horizontal_domain, pseudodistribution, max_width=max_width), params)
     # This efficiently constructs a 2D numpy with shape (N, 2) from a list of 2-element tuples, where N is the number of seed sites.
     # The first and second columns contains the left and right boundaries of the horizontal domains, respectively.
     HIoIs = np.fromiter(itertools.chain.from_iterable(tasks), count=2 * len(seed_sites), dtype=int).reshape(-1, 2)

test/integration/common.py

@@ -3,10 +3,55 @@
 # SPDX-License-Identifier: MIT
 
 
+import pathlib
+from typing import Sequence
+
+from stripepy.IO import Result, ResultFile
+
+
 def matplotlib_avail() -> bool:
     try:
         import matplotlib
     except ImportError:
         return False
 
     return True
+
+
+def _compare_attributes(f1: ResultFile, f2: ResultFile):
+    assert f1.assembly == f2.assembly
+    assert f1.resolution == f2.resolution
+    assert f1.format == f2.format
+    assert f1.normalization == f2.normalization
+    assert f1.chromosomes == f2.chromosomes
+
+
+def _compare_field(f1: ResultFile, f2: ResultFile, chrom: str, field: str, location: str):
+    df1 = f1.get(chrom, field, location)
+    df2 = f2.get(chrom, field, location)
+
+    assert df1.columns == df2.columns
+    assert df1.equals(df2)
+
+
+def _compare_result(f1: ResultFile, f2: ResultFile, chrom: str, location: str):
+    fields = (
+        "pseudodistribution",
+        "all_minimum_points",
+        "persistence_of_all_minimum_points",
+        "all_maximum_points",
+        "persistence_of_all_maximum_points",
+        "stripes",
+    )
+
+    for field in fields:
+        _compare_field(f1, f2, chrom, field, location)
+
+
+def compare_result_files(reference: pathlib.Path, found: pathlib.Path, chroms: Sequence[str]):
+    with ResultFile(reference) as f1, ResultFile(found) as f2:
+        assert chroms in f1.chroms
+        _compare_attributes(f1, f2)
+        for chrom in chroms:
+            _compare_result(f1, f2, chrom, "LT")
+            _compare_result(f1, f2, chrom, "UT")

test/integration/test_stripepy_call.py

@@ -11,7 +11,7 @@
 
 from stripepy import main
 
-from .common import matplotlib_avail
+from .common import compare_result_files, matplotlib_avail
 
 testdir = pathlib.Path(__file__).resolve().parent.parent
 
@@ -33,6 +33,7 @@ def setup_class():
     @staticmethod
     def test_stripepy_call(tmpdir):
         testfile = testdir / "data" / "4DNFI9GMP2J8.mcool"
+        result_file = testdir / "data" / "result_4DNFI9GMP2J8_v2.hdf5"
         resolution = 10_000
 
         chrom_sizes = hictkpy.MultiResFile(testfile).chromosomes()
@@ -58,11 +59,12 @@ def test_stripepy_call(tmpdir):
         outfile = pathlib.Path(tmpdir) / testfile.stem / str(resolution) / "results.hdf5"
 
         assert outfile.is_file()
-        assert h5py.File(outfile).attrs.get("format", "unknown") == "HDF5::StripePy"
+        compare_result_files(result_file, outfile, [chrom for chrom, size in chrom_sizes if size >= chrom_size_cutoff])
 
     @staticmethod
     def test_stripepy_call_with_roi(tmpdir):
         testfile = testdir / "data" / "4DNFI9GMP2J8.mcool"
+        result_file = testdir / "data" / "result_4DNFI9GMP2J8_v2.hdf5"
         resolution = 10_000
 
         chrom_size_cutoff = max(hictkpy.MultiResFile(testfile).chromosomes().values()) - 1
@@ -96,4 +98,4 @@ def test_stripepy_call_with_roi(tmpdir):
         outfile = pathlib.Path(tmpdir) / testfile.stem / str(resolution) / "results.hdf5"
 
         assert outfile.is_file()
-        assert h5py.File(outfile).attrs.get("format", "unknown") == "HDF5::StripePy"
+        compare_result_files(result_file, outfile, [chrom for chrom, size in chrom_sizes if size >= chrom_size_cutoff])