correctly filtering star alleles

ACEnglish · Aug 1, 2024 · 3150ee6 · 3150ee6
1 parent 3f05bf5
commit 3150ee6
Show file tree

Hide file tree

Showing 18 changed files with 77 additions and 1 deletion.
diff --git a/repo_utils/answer_key/bench/bench_starallele/candidate.refine.bed b/repo_utils/answer_key/bench/bench_starallele/candidate.refine.bed
@@ -0,0 +1 @@
+chr1	6004926	6005445
diff --git a/repo_utils/answer_key/bench/bench_starallele/fn.vcf.gz b/repo_utils/answer_key/bench/bench_starallele/fn.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_starallele/fn.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_starallele/fn.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_starallele/fp.vcf.gz b/repo_utils/answer_key/bench/bench_starallele/fp.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_starallele/fp.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_starallele/fp.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_starallele/log.txt b/repo_utils/answer_key/bench/bench_starallele/log.txt
@@ -0,0 +1,50 @@
+2024-08-01 19:48:18,699 [INFO] Truvari v4.3.0
+2024-08-01 19:48:18,699 [INFO] Command /Users/english/code/truvari/truvari/__main__.py bench -b repo_utils/test_files/variants/star.base.vcf.gz -c repo_utils/test_files/variants/star.comp.vcf.gz -s 0 -o test_results/bench_starallele/
+2024-08-01 19:48:18,700 [INFO] Params:
+{
+    "base": "/Users/english/code/truvari/repo_utils/test_files/variants/star.base.vcf.gz",
+    "comp": "/Users/english/code/truvari/repo_utils/test_files/variants/star.comp.vcf.gz",
+    "output": "test_results/bench_starallele/",
+    "includebed": null,
+    "extend": 0,
+    "debug": false,
+    "reference": null,
+    "refdist": 500,
+    "pctseq": 0.7,
+    "minhaplen": 50,
+    "pctsize": 0.7,
+    "pctovl": 0.0,
+    "typeignore": false,
+    "chunksize": 1000,
+    "bSample": "HG002",
+    "cSample": "HG002",
+    "dup_to_ins": false,
+    "sizemin": 0,
+    "sizefilt": 0,
+    "sizemax": 50000,
+    "passonly": false,
+    "no_ref": false,
+    "pick": "single",
+    "check_monref": true,
+    "check_multi": true
+}
+2024-08-01 19:48:18,705 [INFO] Zipped 7 variants Counter({'base': 6, 'comp': 1})
+2024-08-01 19:48:18,705 [INFO] 1 chunks of 7 variants Counter({'base': 5, '__filtered': 1, 'comp': 1})
+2024-08-01 19:48:18,721 [INFO] Stats: {
+    "TP-base": 0,
+    "TP-comp": 0,
+    "FP": 1,
+    "FN": 5,
+    "precision": 0.0,
+    "recall": 0.0,
+    "f1": null,
+    "base cnt": 5,
+    "comp cnt": 1,
+    "TP-comp_TP-gt": 0,
+    "TP-comp_FP-gt": 0,
+    "TP-base_TP-gt": 0,
+    "TP-base_FP-gt": 0,
+    "gt_concordance": 0,
+    "gt_matrix": {}
+}
+2024-08-01 19:48:18,721 [INFO] Finished bench
diff --git a/repo_utils/answer_key/bench/bench_starallele/params.json b/repo_utils/answer_key/bench/bench_starallele/params.json
@@ -0,0 +1 @@
+{"base": "/Users/english/code/truvari/repo_utils/test_files/variants/star.base.vcf.gz", "comp": "/Users/english/code/truvari/repo_utils/test_files/variants/star.comp.vcf.gz", "output": "test_results/bench_starallele/", "includebed": null, "extend": 0, "debug": false, "reference": null, "refdist": 500, "pctseq": 0.7, "minhaplen": 50, "pctsize": 0.7, "pctovl": 0.0, "typeignore": false, "chunksize": 1000, "bSample": "HG002", "cSample": "HG002", "dup_to_ins": false, "sizemin": 0, "sizefilt": 0, "sizemax": 50000, "passonly": false, "no_ref": false, "pick": "single", "check_monref": true, "check_multi": true}
diff --git a/repo_utils/answer_key/bench/bench_starallele/summary.json b/repo_utils/answer_key/bench/bench_starallele/summary.json
@@ -0,0 +1,17 @@
+{
+    "TP-base": 0,
+    "TP-comp": 0,
+    "FP": 1,
+    "FN": 5,
+    "precision": 0.0,
+    "recall": 0.0,
+    "f1": null,
+    "base cnt": 5,
+    "comp cnt": 1,
+    "TP-comp_TP-gt": 0,
+    "TP-comp_FP-gt": 0,
+    "TP-base_TP-gt": 0,
+    "TP-base_FP-gt": 0,
+    "gt_concordance": 0,
+    "gt_matrix": {}
+}
diff --git a/repo_utils/answer_key/bench/bench_starallele/tp-base.vcf.gz b/repo_utils/answer_key/bench/bench_starallele/tp-base.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_starallele/tp-base.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_starallele/tp-base.vcf.gz.tbi
diff --git a/repo_utils/answer_key/bench/bench_starallele/tp-comp.vcf.gz b/repo_utils/answer_key/bench/bench_starallele/tp-comp.vcf.gz
diff --git a/repo_utils/answer_key/bench/bench_starallele/tp-comp.vcf.gz.tbi b/repo_utils/answer_key/bench/bench_starallele/tp-comp.vcf.gz.tbi
diff --git a/repo_utils/sub_tests/bench.sh b/repo_utils/sub_tests/bench.sh
@@ -78,3 +78,10 @@ run test_bench_badparams $truv bench -b nofile.vcf -c nofile.aga -f notref.fa -o
 if [ $test_bench_badparams ]; then
     assert_exit_code 100
 fi
+
+run test_bench_starallele $truv bench -b $INDIR/variants/star.base.vcf.gz \
+                                      -c $INDIR/variants/star.comp.vcf.gz \
+                                      -s 0 -o $OD/bench_starallele/
+if [ $test_bench_starallele ]; then
+    bench_assert _starallele
+fi
diff --git a/repo_utils/test_files/variants/star.base.vcf.gz b/repo_utils/test_files/variants/star.base.vcf.gz
diff --git a/repo_utils/test_files/variants/star.base.vcf.gz.tbi b/repo_utils/test_files/variants/star.base.vcf.gz.tbi
diff --git a/repo_utils/test_files/variants/star.comp.vcf.gz b/repo_utils/test_files/variants/star.comp.vcf.gz
diff --git a/repo_utils/test_files/variants/star.comp.vcf.gz.tbi b/repo_utils/test_files/variants/star.comp.vcf.gz.tbi
diff --git a/truvari/matching.py b/truvari/matching.py
@@ -155,7 +155,7 @@ def filter_call(self, entry, base=False):
         Returns True if the call should be filtered
         Base has different filtering requirements, so let the method know
         """
-        if self.params.check_monref and entry.alts in (None, '*'):  # ignore monomorphic reference
+        if self.params.check_monref and entry.alts and entry.alts[0] in (None, '*'):  # ignore monomorphic reference
             return True
 
         if self.params.check_multi and len(entry.alts) > 1: