Contig stitcher: implement Frankenstein cut_reference

micall/core/contig_stitcher.py

@@ -54,6 +54,8 @@ class FrankensteinContig(AlignedContig):
     Yet its .seq string looks like a real contig. """
 
     def __init__(self, parts: List[GenotypedContig]):
+        assert len(parts) > 0, "Empty Frankensteins do not exist"
+
         self.parts = [subpart for part in parts for subpart in
                       (part.parts if isinstance(part, FrankensteinContig) else [part])]
 
@@ -75,6 +77,17 @@ def seq(self):
         return ''.join(map(lambda part: part.seq, self.parts))
 
 
+    def cut_reference(self, cut_point: float) -> Tuple['AlignedContig', 'AlignedContig']:
+        last_part = self.parts[-1]
+        left_parts = self.parts[-1:]
+
+        left_remainder, left_overlap = left.cut_reference(right.alignment.r_st - 0.5)
+        right_overlap, right_remainder = right.cut_reference(left.alignment.r_ei + 0.5)
+
+        alignment_left, alignment_right = self.alignment.cut_reference(cut_point)
+        return (AlignedContig(self.contig, alignment_left),
+                AlignedContig(self.contig, alignment_right))
+
 
 def align_to_reference(contig: GenotypedContig):
     aligner = Aligner(seq=contig.ref_seq, preset='map-ont')
@@ -87,7 +100,7 @@ def align_to_reference(contig: GenotypedContig):
     return AlignedContig(contig=contig, alignment=single_cigar_hit)
 
 
-def align_equal(seq1, seq2) -> Tuple[str, str]:
+def align_equal(seq1: str, seq2: str) -> Tuple[str, str]:
     gap_open_penalty = 15
     gap_extend_penalty = 3
     use_terminal_gap_penalty = 1

micall/tests/test_contig_stitcher.py

@@ -59,5 +59,10 @@ def test_3():
         ]
 
     result = list(stitch_contigs(contigs))
-    assert 100 == sum(len(x.seq) for x in result)
-    assert result[0].contig.name == 'a+overlap(a,b)+b'
+    assert len(result) == 1
+
+    result = result[0]
+
+    assert 100 == len(result.seq)
+    assert result.seq == 'A' * 50 + 'C' * 50
+    assert result.contig.name == 'a+overlap(a,b)+b'