test coverage of collapse --chain for #196

A single site, but it covers the logic

repo_utils/answer_key/collapse/inputissue196_collapsed.vcf

@@ -0,0 +1,12 @@
+##fileformat=VCFv4.2
+##FILTER=<ID=PASS,Description="All filters passed">
+##contig=<ID=chr21,length=46709983,md5=2f45a3455007b7e271509161e52954a9>
+##INFO=<ID=END,Number=1,Type=Integer,Description="SV END position">
+##INFO=<ID=SVTYPE,Number=A,Type=String,Description="Variant type">
+##INFO=<ID=SVLEN,Number=.,Type=Integer,Description="Difference in length of ref and alt alleles">
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+##INFO=<ID=NumCollapsed,Number=1,Type=Integer,Description="Number of calls collapsed into this call by truvari">
+##INFO=<ID=CollapseId,Number=1,Type=String,Description="Truvari uid to help tie output.vcf and output.collapsed.vcf entries together">
+##INFO=<ID=NumConsolidated,Number=1,Type=Integer,Description="Number of samples consolidated into this call by truvari">
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	NA12878
+chr21	14088112	chr21-14088113-DEL-223	A	<DEL>	.	.	END=14088558;SVTYPE=DEL;SVLEN=-223;NumCollapsed=2;NumConsolidated=0;CollapseId=1.0	GT	1|0

repo_utils/answer_key/collapse/inputissue196_removed.vcf

@@ -0,0 +1,20 @@
+##fileformat=VCFv4.2
+##FILTER=<ID=PASS,Description="All filters passed">
+##contig=<ID=chr21,length=46709983,md5=2f45a3455007b7e271509161e52954a9>
+##INFO=<ID=END,Number=1,Type=Integer,Description="SV END position">
+##INFO=<ID=SVTYPE,Number=A,Type=String,Description="Variant type">
+##INFO=<ID=SVLEN,Number=.,Type=Integer,Description="Difference in length of ref and alt alleles">
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+##INFO=<ID=TruScore,Number=1,Type=Integer,Description="Truvari score for similarity of match">
+##INFO=<ID=PctSeqSimilarity,Number=1,Type=Float,Description="Pct sequence similarity between this variant and its closest match">
+##INFO=<ID=PctSizeSimilarity,Number=1,Type=Float,Description="Pct size similarity between this variant and its closest match">
+##INFO=<ID=PctRecOverlap,Number=1,Type=Float,Description="Percent reciprocal overlap percent of the two calls' coordinates">
+##INFO=<ID=StartDistance,Number=1,Type=Integer,Description="Distance of the base call's end from comparison call's start">
+##INFO=<ID=EndDistance,Number=1,Type=Integer,Description="Distance of the base call's end from comparison call's end">
+##INFO=<ID=SizeDiff,Number=1,Type=Float,Description="Difference in size of base and comp calls">
+##INFO=<ID=GTMatch,Number=1,Type=Integer,Description="Base/Comparison genotypes AC difference">
+##INFO=<ID=MatchId,Number=.,Type=String,Description="Tuple of base and comparison call ids which were matched">
+##INFO=<ID=Multi,Number=0,Type=Flag,Description="Call is false due to non-multimatching">
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	NA12878
+chr21	14088212	chr21-14088113-DEL-223	T	<DEL>	.	.	END=14088828;SVTYPE=DEL;SVLEN=-393;PctSeqSimilarity=0;PctSizeSimilarity=0.5674;PctRecOverlap=0.5624;SizeDiff=-170;StartDistance=-100;EndDistance=-270;GTMatch=.;TruScore=37;MatchId=1.0	GT	1|0
+chr21	14088312	chr21-14088113-DEL-223	A	<DEL>	.	.	END=14089203;SVTYPE=DEL;SVLEN=-668;PctSeqSimilarity=0;PctSizeSimilarity=0.5883;PctRecOverlap=0.5796;SizeDiff=275;StartDistance=100;EndDistance=375;GTMatch=.;TruScore=38;MatchId=1.0	GT	1|0

repo_utils/sub_tests/collapse.sh

@@ -98,3 +98,13 @@ run test_collapse_intragth $truv collapse -i $INDIR/variants/bcftools_merged.vcf
 if [ $test_collapse_intragth ]; then
     collapse_assert intragth
 fi
+
+
+run test_collapse_chain $truv collapse -i $INDIR/variants/issue196_chain.vcf.gz \
+                        -o $OD/inputissue196_collapsed.vcf \
+                        -c $OD/inputissue196_removed.vcf \
+                        --chain --pctseq 0 --pctsize 0.35
+if [ $test_collapse_chain ]; then
+    collapse_assert issue196
+fi
+

truvari/collapse.py

@@ -31,14 +31,7 @@ class CollapsedCalls():
     match_id: str
     matches: list = field(default_factory=list)
     gt_consolidate_count: int = 0
-    genotype_mask: str = ""  # bad
-
-    def combine(self, other):
-        """
-        Put other's entries into this' collapsed_entries
-        """
-        self.matches.append(other.entry)
-        self.matches.extend(other.matches)
+    genotype_mask: str = None # not actually a str
 
     def calc_median_sizepos(self):
         """
@@ -94,29 +87,33 @@ def gt_conflict(self, other, which_gt):
         self.genotype_mask |= o_mask
         return False
 
-    def consolidate(self, other):
+    def combine(self, other):
         """
         Add other's calls/matches to self's matches
         """
-        self.matches.append(other.entry)
         self.matches.extend(other.matches)
         self.gt_consolidate_count += other.gt_consolidate_count
-        if self.genotype_mask != "" and other.genotype_mask != "":
+        if self.genotype_mask is not None and other.genotype_mask is not None:
             self.genotype_mask |= other.genotype_mask
 
 def chain_collapse(cur_collapse, all_collapse, matcher):
     """
     Perform transitive matching of cur_collapse to all_collapse
+    Check the cur_collapse's entry to all other collapses' consolidated entries
     """
     for m_collap in all_collapse:
         for other in m_collap.matches:
             mat = matcher.build_match(cur_collapse.entry,
-                                      other.base,
+                                      other.comp,
                                       m_collap.match_id,
                                       skip_gt=True,
                                       short_circuit=True)
             if mat.state:
-                m_collap.consolidate(cur_collapse)
+                # The other's representative entry will report its
+                # similarity to the matched call that pulled it in
+                mat.base, mat.comp = mat.comp, mat.base
+                m_collap.matches.append(mat)
+                m_collap.combine(cur_collapse)
                 return True  # you can just ignore it later
     return False  # You'll have to add it to your set of collapsed calls