From fab2aa02842145b411d73c1957674b6c759f2b1a Mon Sep 17 00:00:00 2001
From: Roberto Rossini <71787608+robomics@users.noreply.github.com>
Date: Mon, 13 Jan 2025 13:37:37 +0100
Subject: [PATCH] Update tests

---
 test/integration/common.py             | 54 ++++++++++++++++++++++++++
 test/integration/test_stripepy_call.py | 19 +++++----
 2 files changed, 66 insertions(+), 7 deletions(-)

diff --git a/test/integration/common.py b/test/integration/common.py
index 4a5baa1..078e82a 100644
--- a/test/integration/common.py
+++ b/test/integration/common.py
@@ -3,6 +3,14 @@
 # SPDX-License-Identifier: MIT
 
 
+import pathlib
+from typing import Sequence
+
+from pandas.testing import assert_frame_equal
+
+from stripepy.IO import ResultFile
+
+
 def matplotlib_avail() -> bool:
     try:
         import matplotlib
@@ -10,3 +18,49 @@ def matplotlib_avail() -> bool:
         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
+
+    metadata1 = f1.metadata
+    metadata2 = f2.metadata
+
+    metadata1.pop("min-chromosome-size")
+    metadata2.pop("min-chromosome-size")
+
+    assert metadata1 == metadata2
+
+
+def _compare_field(f1: ResultFile, f2: ResultFile, chrom: str, field: str, location: str):
+    assert chrom in f1.chromosomes
+    df1 = f1.get(chrom, field, location)
+    df2 = f2.get(chrom, field, location)
+
+    assert_frame_equal(df1, df2, check_exact=False)
+
+
+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:
+        _compare_attributes(f1, f2)
+        for chrom in chroms:
+            _compare_result(f1, f2, chrom, "LT")
+            _compare_result(f1, f2, chrom, "UT")
diff --git a/test/integration/test_stripepy_call.py b/test/integration/test_stripepy_call.py
index b915d9f..3162c69 100644
--- a/test/integration/test_stripepy_call.py
+++ b/test/integration/test_stripepy_call.py
@@ -5,13 +5,12 @@
 import pathlib
 import warnings
 
-import h5py
 import hictkpy
 import pytest
 
 from stripepy import main
 
-from .common import matplotlib_avail
+from .common import compare_result_files, matplotlib_avail
 
 testdir = pathlib.Path(__file__).resolve().parent.parent
 
@@ -22,6 +21,7 @@ class TestStripePyCall:
     def setup_class():
         test_files = [
             testdir / "data" / "4DNFI9GMP2J8.mcool",
+            testdir / "data" / "results_4DNFI9GMP2J8_v2.hdf5",
         ]
 
         for f in test_files:
@@ -33,6 +33,7 @@ def setup_class():
     @staticmethod
     def test_stripepy_call(tmpdir):
         testfile = testdir / "data" / "4DNFI9GMP2J8.mcool"
+        result_file = testdir / "data" / "results_4DNFI9GMP2J8_v2.hdf5"
         resolution = 10_000
 
         chrom_sizes = hictkpy.MultiResFile(testfile).chromosomes()
@@ -43,7 +44,7 @@ def test_stripepy_call(tmpdir):
             str(testfile),
             str(resolution),
             "--glob-pers-min",
-            "0.10",
+            "0.05",
             "--loc-pers-min",
             "0.33",
             "--loc-trend-min",
@@ -58,21 +59,25 @@ 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.items() if size >= chrom_size_cutoff]
+        )
 
     @staticmethod
     def test_stripepy_call_with_roi(tmpdir):
         testfile = testdir / "data" / "4DNFI9GMP2J8.mcool"
+        result_file = testdir / "data" / "results_4DNFI9GMP2J8_v2.hdf5"
         resolution = 10_000
 
-        chrom_size_cutoff = max(hictkpy.MultiResFile(testfile).chromosomes().values()) - 1
+        chrom_sizes = hictkpy.MultiResFile(testfile).chromosomes()
+        chrom_size_cutoff = max(chrom_sizes.values()) - 1
 
         args = [
             "call",
             str(testfile),
             str(resolution),
             "--glob-pers-min",
-            "0.10",
+            "0.05",
             "--loc-pers-min",
             "0.33",
             "--loc-trend-min",
@@ -96,4 +101,4 @@ 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, [tuple(chrom_sizes.keys())[0]])