varcode 9.2.2__tar.gz → 9.2.3__tar.gz

{varcode-9.2.2/varcode.egg-info → varcode-9.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: varcode
-Version: 9.2.2
+Version: 9.2.3
 Summary: Variant annotation in Python
 Author-email: Alex Rubinsteyn <alex.rubinsteyn@unc.edu>
 Project-URL: Homepage, https://github.com/openvax/varcode

varcode-9.2.3/tests/test_nonlocal_sv_transcripts.py

@@ -0,0 +1,83 @@
+"""A partial overlap cannot establish a local DUP/INV transcript (#405)."""
+
+import pytest
+from pyensembl import cached_release
+
+from varcode import StructuralVariant, parse_symbolic_alt
+from varcode.effects import Inversion, LargeDuplication
+from varcode.effects import structural
+
+
+@pytest.fixture(params=["ENST00000003084", "ENST00000357654"])
+def transcript(request):
+    return cached_release(81).transcript_by_id(request.param)
+
+
+@pytest.fixture(autouse=True)
+def no_optional_candidates(monkeypatch):
+    # Isolate the span fallback; fusion preservation also has real esvee tests.
+    monkeypatch.setattr(structural, "_fusion_partner", lambda *args: None)
+    monkeypatch.setattr(structural, "_enumerate_and_attach_cryptics", lambda *args: None)
+    monkeypatch.setattr(structural, "_enumerate_and_attach_splice_outcomes", lambda *args: None)
+
+
+def make_sv(transcript, kind, outside, symbolic=False, assembly=None):
+    start, end = transcript.start + 50, transcript.end - 50
+    if outside in ("left", "both"):
+        start = transcript.start - 100
+    if outside in ("right", "both"):
+        end = transcript.end + 100
+    if symbolic:
+        return parse_symbolic_alt(
+            transcript.contig, start, "N", "<%s>" % kind,
+            info={"END": end}, genome=transcript.genome)
+    return StructuralVariant(
+        transcript.contig, start, kind, end=end,
+        genome=transcript.genome, alt_assembly=assembly)
+
+
+@pytest.mark.parametrize("kind,cls", [("DUP", LargeDuplication), ("INV", Inversion)])
+@pytest.mark.parametrize("outside", ["left", "right", "both"])
+@pytest.mark.parametrize("symbolic", [False, True])
+def test_nonlocal_event_keeps_type_without_inventing_transcript(
+        transcript, kind, cls, outside, symbolic):
+    sv = make_sv(transcript, kind, outside, symbolic)
+    effect = sv.effect_on_transcript(transcript)
+    assert isinstance(effect, cls)
+    assert effect.mutant_transcript is None
+    assert all(c.effect.mutant_transcript is None for c in effect.candidates)
+
+
+@pytest.mark.parametrize("kind", ["DUP", "INV"])
+@pytest.mark.parametrize("outside", ["left", "right", "both"])
+def test_nonlocal_event_preserves_supplied_assembly(transcript, kind, outside):
+    assembly = "ATG" + "GCT" * 10 + "TAA"
+    sv = make_sv(transcript, kind, outside, assembly=assembly)
+    effect = sv.effect_on_transcript(transcript)
+    assert effect.mutant_transcript.cdna_sequence == assembly
+
+
+@pytest.mark.parametrize("kind", ["DUP", "INV"])
+def test_wholly_internal_event_retains_local_model(transcript, kind):
+    sv = make_sv(transcript, kind, outside=None)
+    effect = sv.effect_on_transcript(transcript)
+    assert effect.mutant_transcript is not None
+    if kind == "DUP":
+        assert len(effect.mutant_transcript.cdna_sequence) > len(transcript.sequence)
+    else:
+        assert any(s.strand == "-" for s in effect.mutant_transcript.reference_segments)
+
+
+@pytest.mark.parametrize("transcript_id", ["ENST00000439541", "ENST00000550411"])
+@pytest.mark.parametrize("kind", ["DUP", "INV"])
+def test_gene_containment_does_not_establish_short_isoform_structure(transcript_id, kind):
+    # CBX5's long isoform spans both ends, but these short isoforms do not.
+    genome = cached_release(95)
+    tx = genome.transcript_by_id(transcript_id)
+    sv = StructuralVariant("12", 54_259_010, kind, end=54_262_307, genome=genome)
+    ends = [end for junction in sv.junctions for end in junction]
+    assert all(structural._contains(tx.gene, end) for end in ends)
+    assert any(not structural._contains(tx, end) for end in ends)
+    effect = sv.effect_on_transcript(tx)
+    assert isinstance(effect, (LargeDuplication, Inversion))
+    assert effect.mutant_transcript is None

{varcode-9.2.2 → varcode-9.2.3}/tests/test_osteosarc_fusions.py

@@ -48,7 +48,7 @@ from varcode import StructuralVariant, FastEffectAnnotator, load_vcf
 from varcode.effects.structural import (
     _build_fusion_mutant_transcript,
 )
-from varcode.effects import GeneFusion, Intronic
+from varcode.effects import GeneFusion, Intronic, Inversion, LargeDuplication
 from varcode.transforms import pair_breakends
 
 from .data import data_path
@@ -267,6 +267,10 @@ def test_esvee_vcf_otx1_kif3c_inversion_is_fusion_on_both_partners():
             assert fusion.three_prime_transcript.id == kif3c.id
             assert fusion.mutant_transcript.mutant_protein_sequence == (
                 "MMSYLKQPPYGMNGLGLAGPAMDLLHPSVGYPETS")
+            if sv.sv_type == "INV":
+                (span,) = [c.effect for c in fusion.candidates
+                           if isinstance(c.effect, Inversion)]
+                assert span.mutant_transcript is None
 
 
 def test_esvee_vcf_cpeb2_fam193a_duplication_is_fusion():
@@ -290,6 +294,10 @@ def test_esvee_vcf_cpeb2_fam193a_duplication_is_fusion():
     assert isinstance(on_fam193a, GeneFusion)
     assert on_fam193a.three_prime_transcript.id == fam193a.id
     assert on_fam193a.five_prime_transcript.gene_name == "CPEB2"
+    for fusion in (on_cpeb2, on_fam193a):
+        (span,) = [c.effect for c in fusion.candidates
+                   if isinstance(c.effect, LargeDuplication)]
+        assert span.mutant_transcript is None
 
 
 def test_esvee_vcf_sema6a_parm1_fusion_from_either_record():

{varcode-9.2.2 → varcode-9.2.3}/tests/test_symbolic_sv_padding.py

@@ -37,7 +37,7 @@ def test_no_nonempty_span_cannot_delete_or_copy_the_anchor(kind, info):
 def test_exact_mutant_cdna_on_both_strands(kind, builder, reverse):
     sequence = "ACGTTGCAAGCTTAGGCTAC"
     tx = SimpleNamespace(exons=[SimpleNamespace(start=100, end=119)],
-                         start=100, end=119, sequence=sequence,
+                         contig="1", start=100, end=119, sequence=sequence,
                          on_backward_strand=reverse)
     sv = parse_symbolic_alt("1", 109, "G", "<%s>" % kind, info={"END": 114})
     a, b = (5, 10) if reverse else (10, 15)

{varcode-9.2.2 → varcode-9.2.3}/varcode/effects/structural.py

@@ -126,9 +126,15 @@ def _cdna_ranges_kept_after_deletion(variant, transcript):
 
 
 def _cdna_ranges_within_sv(variant, transcript):
-    """cDNA ranges that fall INSIDE the SV span (used for DUP to
-    derive the duplicated body and for INV to derive the flipped
-    middle)."""
+    """Body ranges for a local DUP/INV model, or [] if not local.
+
+    Clipping a cross-boundary event to this transcript invents a local
+    rearrangement (#405). Every junction end must be in this transcript,
+    not merely in its gene. Fusion and supplied-assembly paths are separate.
+    """
+    if any(not _contains(transcript, end)
+           for junction in variant.junctions for end in junction):
+        return []
     sv_start, sv_end = _affected_span(variant)
     inside = []
     reverse = transcript.on_backward_strand

varcode-9.2.3/varcode/version.py

@@ -0,0 +1 @@
+__version__ = "9.2.3"

{varcode-9.2.2 → varcode-9.2.3/varcode.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: varcode
-Version: 9.2.2
+Version: 9.2.3
 Summary: Variant annotation in Python
 Author-email: Alex Rubinsteyn <alex.rubinsteyn@unc.edu>
 Project-URL: Homepage, https://github.com/openvax/varcode

{varcode-9.2.2 → varcode-9.2.3}/varcode.egg-info/SOURCES.txt

@@ -45,6 +45,7 @@ tests/test_mouse.py
 tests/test_mutant_transcript.py
 tests/test_mutate.py
 tests/test_no_duplicate_variants.py
+tests/test_nonlocal_sv_transcripts.py
 tests/test_osteosarc_fusions.py
 tests/test_partial_annotators.py
 tests/test_premature_stop_short_description.py

varcode-9.2.2/varcode/version.py

@@ -1 +0,0 @@
-__version__ = "9.2.2"