fix variant end

Author: Hoeze

kipoiseq/dataclasses.py

@@ -85,6 +85,12 @@ def start(self):
         """
         return self.pos - 1
 
+    @property
+    def end(self):
+        """1-based variant end position
+        """
+        return self.start + len(self.ref)
+
     @classmethod
     def from_cyvcf(cls, obj):
         if len(obj.ALT) > 1:

kipoiseq/extractors/vcf_matching.py

@@ -30,7 +30,7 @@ def variants_to_pyranges(variants: List[Variant]) -> PyRanges:
         (
             v.chrom,
             v.start,
-            v.start + max(len(v.ref), len(v.alt)),
+            v.end,
             v
         )
         for v in variants
@@ -200,7 +200,7 @@ def iter_rows(self):
                 for _, row in df.iterrows():
                     yield row
 
-    def __iter__(self):
+    def __iter__(self) -> (Interval, Variant):
         """
         Iterate interval and variant object.
         """

tests/extractors/test_vcf_matching.py

@@ -4,7 +4,7 @@
 from kipoiseq.dataclasses import Interval, Variant
 from kipoiseq.extractors.vcf import MultiSampleVCF
 from kipoiseq.extractors.vcf_matching import variants_to_pyranges, \
-    pyranges_to_intervals,  intervals_to_pyranges, BaseVariantMatcher, \
+    pyranges_to_intervals, intervals_to_pyranges, BaseVariantMatcher, \
     SingleVariantMatcher, MultiVariantsMatcher
 
 intervals = [
@@ -112,27 +112,24 @@ def test_BaseVariantMatcher__read_intervals():
 def test_SingleVariantMatcher__iter__():
     inters = intervals + [Interval('chr1', 5, 50)]
 
-    matcher = SingleVariantMatcher(
-        vcf_file, intervals=inters)
+    matcher = SingleVariantMatcher(vcf_file, intervals=inters)
     pairs = list(matcher)
 
     assert (inters[0], variants[0]) in pairs
     assert (inters[0], variants[1]) in pairs
     assert (inters[1], variants[2]) in pairs
-    assert (inters[2], variants[1]) in pairs
     assert (inters[2], variants[2]) in pairs
 
-    assert len(pairs) == 5
+    assert len(pairs) == 4
 
     matcher = SingleVariantMatcher(vcf_file, pranges=pr)
     pairs = list(matcher)
 
     assert (inters[0], variants[0]) in pairs
     assert (inters[0], variants[1]) in pairs
     assert (inters[1], variants[2]) in pairs
-    assert (inters[2], variants[1]) in pairs
     assert (inters[2], variants[2]) in pairs
-    assert len(pairs) == 5
+    assert len(pairs) == 4
 
 
 def test_MultiVariantMatcher__iter__():