tweaks

bnd positions in decomposition are corrected around anchor base 4x unroll seqsim to check all possibilities break ties in matches by start/end distance (rare)
ACEnglish · Jan 30, 2025 · 839be09 · 839be09
1 parent 664fd5b
commit 839be09
Show file tree

Hide file tree

Showing 12 changed files with 31 additions and 28 deletions.
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/fn.vcf.gz b/repo_utils/answer_key/bench/bench_bnd_decomp/fn.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/fn.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_bnd_decomp/fn.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/fp.vcf.gz b/repo_utils/answer_key/bench/bench_bnd_decomp/fp.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/fp.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_bnd_decomp/fp.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/log.txt b/repo_utils/answer_key/bench/bench_bnd_decomp/log.txt
@@ -1,6 +1,6 @@
-2025-01-29 01:41:36,655 [INFO] Truvari v5.0.1.dev0+24859cf.uc
-2025-01-29 01:41:36,657 [INFO] Command /data/truvari/__main__.py bench -b repo_utils/test_files/variants/bnd.base.vcf.gz -c repo_utils/test_files/variants/bnd.comp2.vcf.gz --sizemax 1000000000 -p 0 --pick multi -o test_results/bench_bnd_decomp/
-2025-01-29 01:41:36,658 [INFO] Params:
+2025-01-30 21:40:22,568 [INFO] Truvari v5.0.1.dev0+664fd5b
+2025-01-30 21:40:22,569 [INFO] Command /data/truvari/__main__.py bench -b repo_utils/test_files/variants/bnd.base.vcf.gz -c repo_utils/test_files/variants/bnd.comp2.vcf.gz --sizemax 1000000000 -p 0 --pick multi -o test_results/bench_bnd_decomp/
+2025-01-30 21:40:22,570 [INFO] Params:
 {
     "base": "/data/repo_utils/test_files/variants/bnd.base.vcf.gz",
     "comp": "/data/repo_utils/test_files/variants/bnd.comp2.vcf.gz",
@@ -36,10 +36,10 @@
     "skip_gt": false,
     "max_resolve": 25000
 }
-2025-01-29 01:41:36,746 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
-2025-01-29 01:41:58,627 [INFO] Zipped 30098 variants Counter({'comp': 29902, 'base': 196})
-2025-01-29 01:41:58,627 [INFO] 20660 chunks of 30098 variants Counter({'comp': 29902, 'base': 195, '__filtered': 1})
-2025-01-29 01:42:03,703 [INFO] Stats: {
+2025-01-30 21:40:22,652 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
+2025-01-30 21:40:43,051 [INFO] Zipped 30098 variants Counter({'comp': 29902, 'base': 196})
+2025-01-30 21:40:43,051 [INFO] 20660 chunks of 30098 variants Counter({'comp': 29902, 'base': 195, '__filtered': 1})
+2025-01-30 21:40:47,691 [INFO] Stats: {
     "TP-base": 152,
     "TP-comp": 131,
     "FP": 28401,
@@ -79,4 +79,4 @@
         }
     }
 }
-2025-01-29 01:42:03,704 [INFO] Finished bench
+2025-01-30 21:40:47,692 [INFO] Finished bench
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/tp-base.vcf.gz b/repo_utils/answer_key/bench/bench_bnd_decomp/tp-base.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/tp-base.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_bnd_decomp/tp-base.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/tp-comp.vcf.gz b/repo_utils/answer_key/bench/bench_bnd_decomp/tp-comp.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/tp-comp.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_bnd_decomp/tp-comp.vcf.gz.tbi
diff --git a/truvari/comparisons.py b/truvari/comparisons.py
@@ -47,7 +47,10 @@ def best_seqsim(a_seq, b_seq, st_dist):
     else:
         rssm = 0
     return max(rssm, unroll_seqsim(a_seq, b_seq, st_dist),
-               unroll_seqsim(b_seq, a_seq, -st_dist), seqsim(a_seq, b_seq))
+               unroll_seqsim(a_seq, b_seq, -st_dist),
+               unroll_seqsim(b_seq, a_seq, st_dist),
+               unroll_seqsim(b_seq, a_seq, -st_dist),
+               seqsim(a_seq, b_seq))
 
 
 def roll_seqsim(a_seq, b_seq):

diff --git a/truvari/matching.py b/truvari/matching.py
@@ -66,7 +66,7 @@ def __init__(self):
         self.gt_match = None
         self.multi = None
         self.state = False
-        self.score = None
+        self.score = 0
 
     def calc_score(self):
         """
@@ -76,12 +76,15 @@ def calc_score(self):
             self.score = (self.seqsim + self.sizesim + self.ovlpct) / 3.0 * 100
 
     def __lt__(self, other):
-        # Trues are always worth more
-        if self.state != other.state:
-            return self.state < other.state
-        m_score = self.score if self.score is not None else -float('inf')
-        o_score = other.score if other.score is not None else -float('inf')
-        return m_score < o_score
+        def s_abs(value):
+            """
+            Negative because we want them to be closer
+            """
+            return -abs(value) if value is not None else -float('inf')
+        return (
+            (self.state, self.score, s_abs(self.st_dist), s_abs(self.ed_dist)) <
+            (other.state, other.score, s_abs(other.st_dist), s_abs(other.ed_dist))
+        )
 
     def __eq__(self, other):
         return self.state == other.state and self.score == other.score

diff --git a/truvari/variant_record.py b/truvari/variant_record.py
@@ -313,11 +313,12 @@ def decompose(self):
 
         if svtype == truvari.SV.INV:
             record1 = self.copy()
-            record1.alts = (f"[{self.chrom}:{self.end}[N",)
+            record1.alts = (f"N]{self.chrom}:{self.end}]",)
             record1.info["SVTYPE"] = "BND"
 
             record2 = self.copy()
-            record2.alts = (f"N]{self.chrom}:{self.end}]",)
+            record2.pos += 1
+            record2.alts = (f"[{self.chrom}:{self.end + 1}[N",)
             record2.info["SVTYPE"] = "BND"
 
             record3 = self.copy()
@@ -326,19 +327,19 @@ def decompose(self):
             record3.info["SVTYPE"] = "BND"
 
             record4 = self.copy()
-            record4.pos = self.end
-            record4.alts = (f"[{self.chrom}:{self.pos}[N",)
+            record4.pos = self.end + 1
+            record4.alts = (f"[{self.chrom}:{self.pos + 1}[N",)
             record4.info["SVTYPE"] = "BND"
 
             ret = [record1, record2, record3, record4]
 
         elif svtype == truvari.SV.DEL:
             record1 = self.copy()
-            record1.alts = (f"N[{self.chrom}:{self.end}[",)
+            record1.alts = (f"N[{self.chrom}:{self.end + 1}[",)
             record1.info["SVTYPE"] = "BND"
 
             record2 = self.copy()
-            record2.pos = self.end
+            record2.pos = self.end + 1
             record2.alts = (f"]{self.chrom}:{self.pos}]N",)
             record2.info["SVTYPE"] = "BND"
 
@@ -347,12 +348,13 @@ def decompose(self):
         elif svtype == truvari.SV.DUP:
             # Assumes DUP:TANDEM
             record1 = self.copy()
+            record1.pos += 1
             record1.alts = (f"]{self.chrom}:{self.end}]N",)
             record1.info["SVTYPE"] = "BND"
 
             record2 = self.copy()
             record2.pos = self.end
-            record2.alts = (f"N[{self.chrom}:{self.pos}[",)
+            record2.alts = (f"N[{self.chrom}:{self.pos + 1}[",)
             record2.info["SVTYPE"] = "BND"
 
             ret = [record1, record2]
@@ -734,11 +736,6 @@ def seqsim(self, other, roll=True):
         if not roll or st_dist == 0 or ed_dist == 0:
             return truvari.seqsim(a_seq, b_seq)
 
-        if st_dist < 0:
-            st_dist *= -1
-        else:
-            a_seq, b_seq = b_seq, a_seq
-
         # Return best of rolled, unrolled from both ends, and direct similarity
         # Whichever is highest is how similar these sequences can be
         return truvari.best_seqsim(a_seq, b_seq, st_dist)