varcode 11.3.2__tar.gz → 11.3.3__tar.gz

{varcode-11.3.2 → varcode-11.3.3}/CHANGELOG.md

@@ -1,5 +1,15 @@
 # Change Log
 
+## v11.3.3
+
+- Stop warning that a reference is ambiguous when its file name matches only
+  one assembly under its Ensembl and UCSC names, such as `CanFam3.1`,
+  `CanFam3.1.dna.toplevel.fa` or `GRCh37_hg19.fa`. The chosen assembly is
+  unchanged.
+- Assert the warnings tests trigger on purpose (skipped symbolic and REF == ALT
+  alleles, ambiguous reference paths, unoriented HCC1395 translocations), and
+  ignore the once-per-genome no-FASTA cryptic-exon warning across the suite.
+
 ## v11.3.2
 
 - Type VCF compression as the literal values pandas accepts, and pin

{varcode-11.3.2/varcode.egg-info → varcode-11.3.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: varcode
-Version: 11.3.2
+Version: 11.3.3
 Summary: Variant annotation in Python
 Author-email: Alex Rubinsteyn <alex.rubinsteyn@unc.edu>
 Project-URL: Homepage, https://github.com/openvax/varcode

{varcode-11.3.2 → varcode-11.3.3}/pyproject.toml

@@ -49,6 +49,12 @@ varcode = "varcode.cli.main:main"
 markers = [
     "parity: fast vs protein_diff byte-for-byte parity harness (tests/test_protein_diff_parity.py). Runs by default; also selectable with `pytest -m parity`.",
 ]
+# Test genomes have no chromosome FASTA, and this warning fires once per
+# genome in whichever test reaches it first. tests/test_genome_sequence.py
+# checks the warning itself.
+filterwarnings = [
+    "ignore:cryptic_exons. no reference sequence:UserWarning",
+]
 
 [tool.ruff.lint]
 select = ["E", "F"]

{varcode-11.3.2 → varcode-11.3.3}/tests/test_genome_sequence.py

@@ -485,27 +485,26 @@ def test_cryptic_exons_warns_once_per_genome():
     # Reset the per-instance flag so the warning is armed.
     g._missing_reference_warned = False
 
+    # No release-81 transcript lies within the 500 bp scan window of
+    # either breakend (the first genes start at 1:11869 and 2:38814).
     sv = StructuralVariant(
-        contig="1", start=1_000_000, sv_type="BND",
-        alt="N]2:1000000]", mate_contig="2",
-        mate_start=1_000_000, mate_orientation="]]",
+        contig="1", start=1_000, sv_type="BND",
+        alt="N]2:1000]", mate_contig="2",
+        mate_start=1_000, mate_orientation="]]",
         genome=g,
     )
 
-    # First call — expect at most one warning (zero is also valid:
-    # transcripts may cover the position and Tier 2 returns content).
     with warnings.catch_warnings(record=True) as first:
         warnings.simplefilter("always")
         enumerate_from_structural_variant(sv, genome=g)
     first_no_ref = [w for w in first
                     if "no reference sequence" in str(w.message)]
-    # Second call — even if first warned, second should NOT.
+    # A second uncovered call on the same genome is silent.
     with warnings.catch_warnings(record=True) as second:
         warnings.simplefilter("always")
         enumerate_from_structural_variant(sv, genome=g)
     second_no_ref = [w for w in second
                      if "no reference sequence" in str(w.message)]
+    assert len(first_no_ref) == 1, [str(w.message) for w in first]
     assert not second_no_ref, [str(w.message) for w in second]
-    # If the first call warned at all, the dedup flag should now be set.
-    if first_no_ref:
-        assert g._missing_reference_warned is True
+    assert g._missing_reference_warned is True

{varcode-11.3.2 → varcode-11.3.3}/tests/test_public_somatic.py

@@ -98,16 +98,26 @@ def test_every_hcc1395_consensus_sv_loads_and_annotates():
     assert Counter(row["SV_type"] for row in rows) == {
         "DEL": 717, "INS": 551, "DUP": 230, "TRA": 146, "INV": 133, "BND": 11}
     genome = cached_release(93)
-    for row in rows:
-        variant = hcc1395_structural_variant(row, genome)
-        assert variant.info["published_consensus"] == row
-        assert variant.alt_assembly is None
-        if row["SV_type"] in ("BND", "TRA"):
-            assert variant.mate_orientation is None
-            assert variant.mate_start == int(row["Pos2"])
-        effects = variant.effects(raise_on_error=True)
-        assert effects, row["SV_id"]
-        materialize_effects(effects)
+    translocations = set()
+    # The table gives each translocation's partner position but not its
+    # orientation, so annotation says which partner it assumed is 5'.
+    with pytest.warns(UserWarning, match="can't read the breakend orientation") as caught:
+        for row in rows:
+            variant = hcc1395_structural_variant(row, genome)
+            assert variant.info["published_consensus"] == row
+            assert variant.alt_assembly is None
+            if row["SV_type"] in ("BND", "TRA"):
+                assert variant.mate_orientation is None
+                assert variant.mate_start == int(row["Pos2"])
+                translocations.add(variant.short_description)
+            effects = variant.effects(raise_on_error=True)
+            assert effects, row["SV_id"]
+            materialize_effects(effects)
+    assumed = {
+        str(w.message).split(" of ", 1)[1].split(" from ALT", 1)[0]
+        for w in caught.list
+        if "can't read the breakend orientation" in str(w.message)}
+    assert assumed <= translocations
 
 
 def test_published_canine_consequences():

{varcode-11.3.2 → varcode-11.3.3}/tests/test_reference.py

@@ -35,38 +35,61 @@ reference_test_cases = {
     ],
 }
 
+# These resolve with a warning: the first file name names no assembly, and
+# in the second "job36" contains NCBI36's alias "b36".
+AMBIGUOUS_REFERENCE_PATHS = {
+    '##reference=file:///var/lib/cwl/ncbi36/homo_sapiens.d1.vd1.fa':
+        "could not be matched against filename",
+    '##reference=file:///var/lib/cwl/job367935311_index_001zdr/GRCh38.job36.d1.vd1.fa':
+        "More than one reference",
+}
+
 def test_most_recent_assembly():
     eq_(most_recent_assembly_name(['ncbi36', 'grch38']), 'grch38')
     eq_(most_recent_assembly_name(['ncbi36', 'grch38', '37mm']), 'grch38')
     eq_(most_recent_assembly_name(['ncbi36']), 'ncbi36')
     eq_(most_recent_assembly_name(['ncbi36', '35']), 'ncbi36')
 def generate_reference_name_aliases():
-    with warnings.catch_warnings(record=True) as w:
-        for assembly_name, aliases in ensembl_reference_aliases.items():
-            candidate_list = [assembly_name] + list(aliases)
-            for candidate in candidate_list:
-                yield (                
-                    candidate,
-                    assembly_name
-                )
+    for assembly_name, aliases in ensembl_reference_aliases.items():
+        candidate_list = [assembly_name] + list(aliases)
+        for candidate in candidate_list:
+            yield (
+                candidate,
+                assembly_name
+            )
 
 @pytest.mark.parametrize(['candidate', 'assembly_name'], list(generate_reference_name_aliases()))
 def test_infer_reference_name_aliases(candidate, assembly_name):
     eq_(infer_reference_name(candidate), assembly_name)
     
 def generate_reference_name_fasta_filenames():
-    with warnings.catch_warnings(record=True):
-        for assembly_name, aliases in reference_test_cases.items():
-            candidate_list = [assembly_name] + list(aliases)
-            for candidate in candidate_list:
-                yield (
-                    candidate,
-                    assembly_name
-                )
+    for assembly_name, aliases in reference_test_cases.items():
+        candidate_list = [assembly_name] + list(aliases)
+        for candidate in candidate_list:
+            yield (
+                candidate,
+                assembly_name
+            )
 
 @pytest.mark.parametrize(['candidate', 'assembly_name'], list(generate_reference_name_fasta_filenames()))
 def test_reference_name_fasta_filenames(candidate, assembly_name):
-    eq_(infer_reference_name(candidate), assembly_name)
+    if candidate in AMBIGUOUS_REFERENCE_PATHS:
+        # Bypass memoization so the warning fires whatever ran first.
+        with pytest.warns(UserWarning, match=AMBIGUOUS_REFERENCE_PATHS[candidate]):
+            eq_(infer_reference_name.__wrapped__(candidate), assembly_name)
+    else:
+        eq_(infer_reference_name(candidate), assembly_name)
+
+
+@pytest.mark.parametrize(["name", "reference_name"], [
+    ("CanFam3.1", "CanFam3.1"),
+    ("/ref/CanFam3.1.dna.toplevel.fa", "CanFam3.1"),
+    ("GRCh37_hg19.fa", "GRCh37"),
+])
+def test_ensembl_and_ucsc_names_of_one_assembly_are_not_ambiguous(name, reference_name):
+    with warnings.catch_warnings():
+        warnings.simplefilter("error")
+        assert infer_reference_name.__wrapped__(name) == reference_name
 
 
 

{varcode-11.3.2 → varcode-11.3.3}/tests/test_structural_variants.py

@@ -346,7 +346,8 @@ def test_vcf_loader_skips_svs_by_default():
     )
     path = _write_vcf(body)
     try:
-        vc = load_vcf(path, genome="GRCh38")
+        with pytest.warns(UserWarning, match="Skipped 1 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh38")
     finally:
         os.unlink(path)
     # Only the normal SNV survives.

{varcode-11.3.2 → varcode-11.3.3}/tests/test_symbolic_alleles.py

@@ -47,7 +47,8 @@ def test_symbolic_cnv_allele_is_skipped():
         "22\t51179100\tnormal_snv\tA\tG\t100\tPASS\t.\n"
     )
     try:
-        vc = load_vcf(path, genome="GRCh37")
+        with pytest.warns(UserWarning, match="Skipped 1 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh37")
     finally:
         os.unlink(path)
     # The normal SNV should load, the symbolic allele should be skipped.
@@ -61,7 +62,8 @@ def test_symbolic_insertion_allele_is_skipped():
         "22\t51068654\tALU_test\tG\t<INS:ME:ALU>\t100\tPASS\t.\n"
     )
     try:
-        vc = load_vcf(path, genome="GRCh37")
+        with pytest.warns(UserWarning, match="Skipped 1 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh37")
     finally:
         os.unlink(path)
     assert len(vc) == 0, \
@@ -75,7 +77,8 @@ def test_breakend_allele_is_skipped():
         "17\t198983\tnormal\tC\tT\t100\tPASS\t.\n"
     )
     try:
-        vc = load_vcf(path, genome="GRCh37")
+        with pytest.warns(UserWarning, match="Skipped 1 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh37")
     finally:
         os.unlink(path)
     assert len(vc) == 1, \
@@ -89,7 +92,8 @@ def test_multiallelic_with_one_symbolic_and_one_normal_keeps_normal():
         "22\t51179178\tmixed\tA\tG,<CN0>\t100\tPASS\t.\n"
     )
     try:
-        vc = load_vcf(path, genome="GRCh37")
+        with pytest.warns(UserWarning, match="Skipped 1 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh37")
     finally:
         os.unlink(path)
     assert len(vc) == 1, \
@@ -108,7 +112,8 @@ def test_mixed_vcf_with_many_symbolic_alleles_does_not_crash():
         "22\t51179100\tnormal\tA\tG\t100\tPASS\t.\n"
     )
     try:
-        vc = load_vcf(path, genome="GRCh37")
+        with pytest.warns(UserWarning, match="Skipped 3 symbolic/breakend allele"):
+            vc = load_vcf(path, genome="GRCh37")
     finally:
         os.unlink(path)
     assert len(vc) == 1, "Expected 1 variant after filtering symbolic alleles"

{varcode-11.3.2 → varcode-11.3.3}/tests/test_vcf.py

@@ -11,6 +11,7 @@
 # limitations under the License.
 
 import os
+import warnings
 
 import pytest 
 
@@ -28,6 +29,10 @@ RUN_TESTS_REQUIRING_INTERNET = bool(int(
     os.environ.get("RUN_TESTS_REQUIRING_INTERNET", 0)))
 
 HG19_VCF_FILENAME = data_path("somatic_hg19_14muts.vcf")
+# One of its 14 records has REF == ALT, which load_vcf skips with a warning.
+# test_load_vcf_local checks the warning; other tests here just load the file.
+REF_IDENTICAL_SKIP = "Skipped 1 reference-identical ALT allele"
+pytestmark = pytest.mark.filterwarnings("ignore:%s:UserWarning" % REF_IDENTICAL_SKIP)
 HG19_VCF_EXTERNAL_URL = (
     "https://raw.githubusercontent.com/hammerlab/varcode/master/test/data/somatic_hg19_14muts.vcf")
 
@@ -37,25 +42,30 @@ HG19_VCF_EXTERNAL_URL = (
 #   "https://raw.githubusercontent.com/hammerlab/varcode/faster-vcf-parsing/test/data/somatic_hg19_14muts.vcf")
 
 def test_load_vcf_local():
-    variants = load_vcf(HG19_VCF_FILENAME)
+    with pytest.warns(UserWarning, match=REF_IDENTICAL_SKIP):
+        variants = load_vcf(HG19_VCF_FILENAME)
     assert variants.reference_names() == {"GRCh37"}
     assert len(variants) == 13
 
-    variants = load_vcf(HG19_VCF_FILENAME + ".gz")
+    with pytest.warns(UserWarning, match=REF_IDENTICAL_SKIP):
+        variants = load_vcf(HG19_VCF_FILENAME + ".gz")
     assert variants.reference_names() == {"GRCh37"}
     assert len(variants) == 13
 
-    variants = load_vcf("file://%s" % HG19_VCF_FILENAME)
+    with pytest.warns(UserWarning, match=REF_IDENTICAL_SKIP):
+        variants = load_vcf("file://%s" % HG19_VCF_FILENAME)
     assert variants.reference_names() == {"GRCh37"}
     assert len(variants) == 13
 
-    variants = load_vcf("file://%s.gz" % HG19_VCF_FILENAME)
+    with pytest.warns(UserWarning, match=REF_IDENTICAL_SKIP):
+        variants = load_vcf("file://%s.gz" % HG19_VCF_FILENAME)
     assert variants.reference_names() == {"GRCh37"}
     assert len(variants) == 13
 
     # An extra slashe before an absolute path can confuse URL parsing.
     # Test that it can still be opened:
-    variants = load_vcf("/%s" % HG19_VCF_FILENAME)
+    with pytest.warns(UserWarning, match=REF_IDENTICAL_SKIP):
+        variants = load_vcf("/%s" % HG19_VCF_FILENAME)
     assert variants.reference_names() == {"GRCh37"}
     assert len(variants) == 13
 
@@ -141,7 +151,10 @@ def test_vcf_number_entries_duplicates():
     assert len(variants) == 3
 
 def generate_vcf_gene_names():
-    variants = load_vcf(HG19_VCF_FILENAME)
+    # Parameters load at collection, outside pytestmark's filter.
+    with warnings.catch_warnings():
+        warnings.filterwarnings("ignore", REF_IDENTICAL_SKIP, UserWarning)
+        variants = load_vcf(HG19_VCF_FILENAME)
     for variant in variants:
         yield (variants, variant)
 

{varcode-11.3.2 → varcode-11.3.3}/tests/test_vcf_output.py

@@ -10,6 +10,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from contextlib import nullcontext
 import tempfile
 from pathlib import Path
 
@@ -53,6 +54,13 @@ def _merge_metadata_naive(variants):
 
 
 
+# Each has one REF == ALT record, which load_vcf skips with a warning.
+REF_IDENTICAL_FIXTURES = {
+    'somatic_hg19_14muts.space_in_sample_name.vcf',
+    'somatic_hg19_14muts.vcf',
+}
+
+
 def _do_roundtrip_test(filenames, convert_ucsc_to_grch37=False):
 
     def load_fn(filename):
@@ -64,7 +72,11 @@ def _do_roundtrip_test(filenames, convert_ucsc_to_grch37=False):
     def load_variants():
         variant_collections = []
         for filename in filenames:
-            variant_collections.append(load_fn(filename)(data_path(filename)))
+            skipped = (
+                pytest.warns(UserWarning, match="reference-identical ALT")
+                if filename in REF_IDENTICAL_FIXTURES else nullcontext())
+            with skipped:
+                variant_collections.append(load_fn(filename)(data_path(filename)))
         return variant_collections[0].union(*variant_collections[1:])
 
     variants = load_variants()

{varcode-11.3.2 → varcode-11.3.3}/varcode/reference.py

@@ -273,10 +273,16 @@ def infer_reference_name(reference_name_or_path):
     elif len(file_name_matches) > 1:
         # separate logic for >1 vs 1 to give informative warning
         match = choose_best_assembly_name(file_name_matches)
-        warn(
-            ('More than one reference ({}) matches path in header ({}); '
-             'the most recent one ({}) was used.').format(
-                ','.join(file_name_matches), reference_name_or_path, match))
+        # A UCSC name (canFam3 inside CanFam3.1) is the same assembly, so
+        # only matches naming different assemblies are ambiguous.
+        assemblies = {
+            ucsc_to_ensembl_reference_names.get(name, name)
+            for name in file_name_matches}
+        if len(assemblies) > 1:
+            warn(
+                ('More than one reference ({}) matches path in header ({}); '
+                 'the most recent one ({}) was used.').format(
+                    ','.join(file_name_matches), reference_name_or_path, match))
     elif len(full_path_matches) >= 1:
         # combine full-path logic since warning is the same
         match = choose_best_assembly_name(full_path_matches)

varcode-11.3.3/varcode/version.py

@@ -0,0 +1 @@
+__version__ = "11.3.3"

{varcode-11.3.2 → varcode-11.3.3/varcode.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: varcode
-Version: 11.3.2
+Version: 11.3.3
 Summary: Variant annotation in Python
 Author-email: Alex Rubinsteyn <alex.rubinsteyn@unc.edu>
 Project-URL: Homepage, https://github.com/openvax/varcode

varcode-11.3.2/varcode/version.py

@@ -1 +0,0 @@
-__version__ = "11.3.2"