Merge pull request #36 from MuhammedHasan/master

critical bug fix, insertion behavior changed

Author: MuhammedHasan

kipoiseq/extractors/vcf_seq.py

@@ -178,7 +178,7 @@ def _split_overlapping(self, variant_pairs, anchor, which='both'):
         Split the variants hitting the anchor into two
         """
         for ref, alt in variant_pairs:
-            if ref.start < anchor < ref.end or alt.start < anchor < alt.end:
+            if ref.start < anchor < ref.end:
                 mid = anchor - ref.start
                 if which == 'left' or which == 'both':
                     yield ref[:mid], alt[:mid]
@@ -208,7 +208,7 @@ def _downstream_builder(self, down_variants, interval, anchor, istart):
 
         prev = anchor
         for ref, alt in down_variants:
-            if ref.end < istart:
+            if ref.end <= istart:
                 break
             down_sb.append(Interval(interval.chrom, ref.end, prev))
             down_sb.append(alt)
@@ -223,7 +223,7 @@ def _upstream_builder(self, up_variants, interval, anchor, iend):
 
         prev = anchor
         for ref, alt in up_variants:
-            if ref.start > iend:
+            if ref.start >= iend:
                 break
             up_sb.append(Interval(interval.chrom, prev, ref.start))
             up_sb.append(alt)
@@ -241,7 +241,7 @@ def _cut_to_fix_len(self,  down_str, up_str, interval, anchor):
         down_len = anchor - interval.start
         up_len = interval.end - anchor
         down_str = down_str[-down_len:] if down_len else ''
-        up_str = up_str[:up_len] if up_len else ''
+        up_str = up_str[: up_len] if up_len else ''
         return down_str, up_str
 
 

tests/extractors/test_vcf_seq_extractor.py

@@ -81,28 +81,29 @@ def test__split_overlapping(variant_seq_extractor):
     assert splited_pairs[1][0].seq == 'A'
     assert splited_pairs[1][1].seq == ''
 
-    pair = (Sequence(seq='T', start=3, end=4),
+    pair = (Sequence(seq='TT', start=3, end=5),
             Sequence(seq='AAA', start=3, end=6))
-    splited_pairs = list(variant_seq_extractor._split_overlapping([pair], 5))
+    splited_pairs = list(variant_seq_extractor._split_overlapping([pair], 4))
 
     assert splited_pairs[0][0].seq == 'T'
-    assert splited_pairs[0][1].seq == 'AA'
-    assert splited_pairs[1][0].seq == ''
-    assert splited_pairs[1][1].seq == 'A'
+    assert splited_pairs[0][1].seq == 'A'
+    assert splited_pairs[1][0].seq == 'T'
+    assert splited_pairs[1][1].seq == 'AA'
 
 
 def test_extract(variant_seq_extractor):
     variants = list(VCF(vcf_file)())
 
     interval = Interval('chr1', 2, 9)
+
     seq = variant_seq_extractor.extract(interval, variants, anchor=5)
     assert len(seq) == interval.end - interval.start
-    assert seq == 'GCGAACG'
+    assert seq == 'CGAACGT'
 
     interval = Interval('chr1', 2, 9, strand='-')
     seq = variant_seq_extractor.extract(interval, variants, anchor=5)
     assert len(seq) == interval.end - interval.start
-    assert seq == 'CGTTCGC'
+    assert seq == 'ACGTTCG'
 
     interval = Interval('chr1', 4, 14)
     seq = variant_seq_extractor.extract(interval, variants, anchor=7)
@@ -142,7 +143,12 @@ def test_extract(variant_seq_extractor):
     interval = Interval('chr1', 5, 11, strand='+')
     seq = variant_seq_extractor.extract(
         interval, variants, anchor=10, fixed_len=False)
-    assert seq == 'AACGTAA'
+    assert seq == 'ACGTAA'
+
+    interval = Interval('chr1', 0, 3, strand='+')
+    seq = variant_seq_extractor.extract(
+        interval, variants, anchor=10, fixed_len=False)
+    assert seq == 'ACG'
 
 
 @pytest.fixture