Patch to fix pgscatalog-intersect with bim input (#25)

* fix pgscatalog-intersect with bim input

* bump patch

* delete unused fixture target_pvar

* drop unneeded try / except block for KeyErrors

pgscatalog.match/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pgscatalog.match"
-version = "0.2.1"
+version = "0.2.2"
 description = "Tools for matching variants in PGS scoring files and target variant information files"
 authors = ["Benjamin Wingfield <bwingfield@ebi.ac.uk>", "Samuel Lambert <sl925@medschl.cam.ac.uk>", "Laurent Gil <lg10@sanger.ac.uk>"]
 readme = "README.md"

pgscatalog.match/src/pgscatalog/match/cli/intersect_cli.py

@@ -93,24 +93,30 @@ def run_intersect():
     o_tmp_t = []
     target_heap = []
     for path in args.target:
+        # assume inputs always have two extensions (bim.zst / pvar.zst)
+        path = pathlib.Path(path)
+        basename = path.name
+        stem = pathlib.Path(basename.split(".")[0])
+
         logger.info("Reading TARGET variants: {}".format(path))
         pvar = read_var_general(
             path, chrom=None
         )  # essential not to filter target (messes up common line indexing)
 
-        loc_afreq = path.replace(".pvar.zst", ".afreq.gz")
+        loc_afreq = path.parent / stem.with_suffix(".afreq.gz")
         afreq = read_var_general(
             loc_afreq, chrom=None
         )  # essential not to filter target (messes up common line indexing)
 
-        loc_vmiss = path.replace(".pvar.zst", ".vmiss.gz")
+        loc_vmiss = path.parent / stem.with_suffix(".vmiss.gz")
         vmiss = read_var_general(
             loc_vmiss, chrom=None
         )  # essential not to filter target (messes up common line indexing)
 
         for v, freq, miss in zip(pvar, afreq, vmiss):
-            if all([v["ID"], freq["ID"], miss["#ID"]]) is False:
+            if not all([v["ID"], freq["ID"], miss["#ID"]]):
                 raise ValueError("TARGET variant files are not sorted")
+
             count_var_t += 1
             ALTs = v["ALT"].split(",")
             ALT_FREQS = [float(x) for x in freq["ALT_FREQS"].split(",")]

pgscatalog.match/tests/test_intersect_cli.py

@@ -17,16 +17,16 @@ def afreq(request):
     return request.path.parent / "data" / "target.afreq.gz"
 
 
-@pytest.fixture(scope="module")
-def target(request):
-    return request.path.parent / "data" / "target.pvar.zst"
-
-
 @pytest.fixture(scope="module")
 def ref(request):
     return request.path.parent / "data" / "ref.pvar.zst"
 
 
+@pytest.fixture(params=["target.bim.zst", "target.pvar.zst"], scope="module")
+def target(request):
+    return request.path.parent / "data" / request.param
+
+
 def test_intersect_cli(tmp_path_factory, ref, target, afreq, vmiss):
     outdir = tmp_path_factory.mktemp("outdir")