Contig stitcher: remove shortcut handling of rc alignments

Author: Donaim

micall/core/contig_stitcher.py

@@ -685,7 +685,7 @@ def stitch_consensus(contigs: Iterable[GenotypedContig]) -> Iterable[GenotypedCo
     consensus_parts: Dict[GroupRef, List[AlignedContig]] = defaultdict(list)
 
     for contig in contigs:
-        if isinstance(contig, AlignedContig) and contig.strand == "forward":
+        if isinstance(contig, AlignedContig):
             consensus_parts[contig.group_ref].append(contig)
         else:
             yield contig

micall/core/plot_contigs.py

@@ -625,6 +625,8 @@ def record_rstrip(result: AlignedContig, original: AlignedContig):
         elif isinstance(event, events.Stitch):
             record_contig(event.result, [event.left, event.right])
             record_alive(event.result)
+        elif isinstance(event, events.NoOverlap):
+            record_alive(event.contig)
         elif isinstance(event, events.Cut):
             record_contig(event.left, [event.original])
             record_contig(event.right, [event.original])
@@ -634,7 +636,7 @@ def record_rstrip(result: AlignedContig, original: AlignedContig):
             if event.contigs:
                 combine_left_edge[event.result.name] = event.contigs[0].name
                 combine_right_edge[event.result.name] = event.contigs[-1].name
-        elif isinstance(event, (events.IgnoreGap, events.NoOverlap, events.InitialHit)):
+        elif isinstance(event, (events.IgnoreGap, events.InitialHit)):
             pass
         else:
             x: NoReturn = event

micall/tests/data/stitcher_plots/test_multiple_reverse_complement_matches_out_of_order.svg

@@ -978,7 +978,8 @@ def test_multiple_reverse_complement_matches_out_of_order(projects, visualizer):
 
     hxb2_name = 'HIV1-B-FR-K03455-seed'
     ref = projects.getReference(hxb2_name)
-    seq = 'A' * 9 + revcomp(ref[2000:2300]) + 'A' * 9 + revcomp(ref[3000:3300]) + 'T' * 9 + revcomp(ref[4000:4300]) + 'G' * 27
+    ref_part = 'T' * 24 + ref[2000:2600] + 'A' * 9 + ref[3000:3600] + 'T' * 9 + ref[4000:4600] + 'G' * 27
+    seq = revcomp(ref_part)
 
     contigs = [
         GenotypedContig(name='a',
@@ -991,13 +992,13 @@ def test_multiple_reverse_complement_matches_out_of_order(projects, visualizer):
         ]
 
     results = list(stitch_consensus(contigs))
-    assert len(results) == 3
-
-    # Note how parts are stripped because aligned out of order
-    assert results[0].seq == ref[2000:2300]
-    assert results[1].seq == ref[3000:3300]
-    assert results[2].seq == ref[4000:4300]
-
+    assert len(results) == 1
+    assert len(results[0].seq) == len(ref_part)
+    assert results[0].seq == ref_part
+    assert len(results[0].lstrip().seq) == len(ref_part) - 24
+    assert len(results[0].rstrip().seq) == len(ref_part) - 27
+    assert results[0].lstrip().seq == ref_part[24:]
+    assert results[0].rstrip().seq == ref_part[:-27]
     assert len(visualizer().elements) > len(contigs)