Comp BND dropping bug

FP BNDs were being dropped
ACEnglish · Feb 16, 2025 · e1405f7 · e1405f7
1 parent ec2fc3c
commit e1405f7
Show file tree

Hide file tree

Showing 10 changed files with 33 additions and 34 deletions.
diff --git a/repo_utils/answer_key/bench/bench_bnd/fp.vcf.gz b/repo_utils/answer_key/bench/bench_bnd/fp.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_bnd/fp.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_bnd/fp.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_bnd/log.txt b/repo_utils/answer_key/bench/bench_bnd/log.txt
@@ -1,6 +1,6 @@
-2025-01-29 01:41:34,415 [INFO] Truvari v5.0.1.dev0+24859cf.uc
-2025-01-29 01:41:34,416 [INFO] Command /data/truvari/__main__.py bench -b repo_utils/test_files/variants/bnd.base.vcf.gz -c repo_utils/test_files/variants/bnd.comp.vcf.gz -p 0 -o test_results/bench_bnd/ --no-decompose
-2025-01-29 01:41:34,418 [INFO] Params:
+2025-02-16 22:44:26,757 [INFO] Truvari v5.2.1.dev0+ec2fc3c.uc
+2025-02-16 22:44:26,758 [INFO] Command /data/truvari/__main__.py bench -b repo_utils/test_files/variants/bnd.base.vcf.gz -c repo_utils/test_files/variants/bnd.comp.vcf.gz -p 0 -o test_results/bench_bnd/ --no-decompose
+2025-02-16 22:44:26,759 [INFO] Params:
 {
     "base": "/data/repo_utils/test_files/variants/bnd.base.vcf.gz",
     "comp": "/data/repo_utils/test_files/variants/bnd.comp.vcf.gz",
@@ -36,19 +36,19 @@
     "skip_gt": false,
     "max_resolve": 25000
 }
-2025-01-29 01:41:34,484 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
-2025-01-29 01:41:34,766 [INFO] Zipped 439 variants Counter({'comp': 243, 'base': 196})
-2025-01-29 01:41:34,767 [INFO] 226 chunks of 439 variants Counter({'comp': 243, 'base': 195, '__filtered': 1})
-2025-01-29 01:41:34,907 [INFO] Stats: {
+2025-02-16 22:44:26,818 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
+2025-02-16 22:44:27,079 [INFO] Zipped 439 variants Counter({'comp': 243, 'base': 196})
+2025-02-16 22:44:27,080 [INFO] 226 chunks of 439 variants Counter({'comp': 243, 'base': 195, '__filtered': 1})
+2025-02-16 22:44:27,201 [INFO] Stats: {
     "TP-base": 120,
     "TP-comp": 120,
-    "FP": 102,
+    "FP": 123,
     "FN": 75,
-    "precision": 0.5405405405405406,
+    "precision": 0.49382716049382713,
     "recall": 0.6153846153846154,
-    "f1": 0.5755395683453237,
+    "f1": 0.547945205479452,
     "base cnt": 195,
-    "comp cnt": 222,
+    "comp cnt": 243,
     "TP-comp_TP-gt": 100,
     "TP-comp_FP-gt": 20,
     "TP-base_TP-gt": 100,
@@ -68,4 +68,4 @@
         }
     }
 }
-2025-01-29 01:41:34,908 [INFO] Finished bench
+2025-02-16 22:44:27,202 [INFO] Finished bench
diff --git a/repo_utils/answer_key/bench/bench_bnd/summary.json b/repo_utils/answer_key/bench/bench_bnd/summary.json
@@ -1,13 +1,13 @@
 {
     "TP-base": 120,
     "TP-comp": 120,
-    "FP": 102,
+    "FP": 123,
     "FN": 75,
-    "precision": 0.5405405405405406,
+    "precision": 0.49382716049382713,
     "recall": 0.6153846153846154,
-    "f1": 0.5755395683453237,
+    "f1": 0.547945205479452,
     "base cnt": 195,
-    "comp cnt": 222,
+    "comp cnt": 243,
     "TP-comp_TP-gt": 100,
     "TP-comp_FP-gt": 20,
     "TP-base_TP-gt": 100,

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 @@
-2025-01-30 22:47:16,962 [INFO] Truvari v5.0.1.dev0+664fd5b
-2025-01-30 22:47:16,963 [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 22:47:16,964 [INFO] Params:
+2025-02-16 22:44:28,882 [INFO] Truvari v5.2.1.dev0+ec2fc3c.uc
+2025-02-16 22:44:28,883 [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/
+2025-02-16 22:44:28,884 [INFO] Params:
 {
     "base": "/data/repo_utils/test_files/variants/bnd.base.vcf.gz",
     "comp": "/data/repo_utils/test_files/variants/bnd.comp2.vcf.gz",
@@ -36,19 +36,19 @@
     "skip_gt": false,
     "max_resolve": 25000
 }
-2025-01-30 22:47:17,047 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
-2025-01-30 22:47:38,251 [INFO] Zipped 30098 variants Counter({'comp': 29902, 'base': 196})
-2025-01-30 22:47:38,252 [INFO] 20660 chunks of 30098 variants Counter({'comp': 29902, 'base': 195, '__filtered': 1})
-2025-01-30 22:47:42,922 [INFO] Stats: {
+2025-02-16 22:44:28,971 [WARNING] 193 contigs present in comparison VCF header are not in baseline VCF.
+2025-02-16 22:44:50,577 [INFO] Zipped 30098 variants Counter({'comp': 29902, 'base': 196})
+2025-02-16 22:44:50,577 [INFO] 20660 chunks of 30098 variants Counter({'comp': 29902, 'base': 195, '__filtered': 1})
+2025-02-16 22:44:55,998 [INFO] Stats: {
     "TP-base": 152,
     "TP-comp": 131,
-    "FP": 28401,
+    "FP": 29558,
     "FN": 43,
-    "precision": 0.004591336043740362,
+    "precision": 0.004412408636195224,
     "recall": 0.7794871794871795,
-    "f1": 0.009128901026932595,
+    "f1": 0.008775144201329448,
     "base cnt": 195,
-    "comp cnt": 28532,
+    "comp cnt": 29689,
     "TP-comp_TP-gt": 1,
     "TP-comp_FP-gt": 130,
     "TP-base_TP-gt": 1,
@@ -79,4 +79,4 @@
         }
     }
 }
-2025-01-30 22:47:42,923 [INFO] Finished bench
+2025-02-16 22:44:55,999 [INFO] Finished bench
diff --git a/repo_utils/answer_key/bench/bench_bnd_decomp/summary.json b/repo_utils/answer_key/bench/bench_bnd_decomp/summary.json
@@ -1,13 +1,13 @@
 {
     "TP-base": 152,
     "TP-comp": 131,
-    "FP": 28401,
+    "FP": 29558,
     "FN": 43,
-    "precision": 0.004591336043740362,
+    "precision": 0.004412408636195224,
     "recall": 0.7794871794871795,
-    "f1": 0.009128901026932595,
+    "f1": 0.008775144201329448,
     "base cnt": 195,
-    "comp cnt": 28532,
+    "comp cnt": 29689,
     "TP-comp_TP-gt": 1,
     "TP-comp_FP-gt": 130,
     "TP-base_TP-gt": 1,

diff --git a/truvari/bench.py b/truvari/bench.py
@@ -385,7 +385,7 @@ def write_match(self, match):
                     box["TP-comp_TP-gt"] += 1
                 else:
                     box["TP-comp_FP-gt"] += 1
-            elif match.comp.var_size() >= self.m_params.sizemin:
+            elif match.comp.var_size() >= self.m_params.sizemin or match.comp.is_bnd():
                 # The if is because we don't count FPs between sizefilt-sizemin
                 box["comp cnt"] += 1
                 box["FP"] += 1

diff --git a/truvari/matching.py b/truvari/matching.py
@@ -166,7 +166,6 @@ def chunker(params, *files):
     cur_end = 0
     cur_chunk = defaultdict(list)
     reference = pysam.FastaFile(params.reference) if params.reference is not None else None
-
     for key, entry in file_zipper(*files):
         if entry.filter_call(key == 'base'):
             cur_chunk['__filtered'].append(entry)