Fix when INFO/tags is a flag #307

bioinfo-chru-strasbourg · Dec 12, 2024 · 0d8589c · 0d8589c
1 parent f6bd475
commit 0d8589c
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 8 deletions.
diff --git a/.gitignore b/.gitignore
@@ -15,6 +15,7 @@ tests/databases/hg19.fa
 tests/databases/hg19.fa.fai
 tests/databases/.DS_Store
 tests/data/TEM*
+tests/*.DS_Store
 benchmark
 misc/
 *.DS_Store

diff --git a/howard/objects/variants.py b/howard/objects/variants.py
@@ -11686,7 +11686,7 @@ def rename_info_fields(
         regex_replace_dict = {}
         regex_replace_nb = 0
         regex_replace_partition = 125
-        regex_replace = "INFO"
+        regex_replace = "concat(INFO, ';')" # Add ';' to reduce regexp comlexity
 
         if fields_to_rename is not None and access not in ["RO"]:
 
@@ -11713,17 +11713,17 @@ def rename_info_fields(
                     del header.infos[field_to_rename]
 
                     # Rename INFO patterns
-                    field_pattern = rf'(^|;)({field_to_rename})($|;|=[^;]*)'
+                    field_pattern = rf'(^|;)({field_to_rename})(=[^;]*)?;'
                     if field_renamed is not None:
-                        field_renamed_pattern = rf'\1{field_renamed}\3'
+                        field_renamed_pattern = rf'\1{field_renamed}\3;'
                     else:
-                        field_renamed_pattern = ''
+                        field_renamed_pattern = r'\1'
 
                     # regexp replace
                     regex_replace_nb += 1
                     regex_replace_key = math.floor(regex_replace_nb / regex_replace_partition)
                     if (regex_replace_nb % regex_replace_partition) == 0:
-                        regex_replace = "INFO"
+                        regex_replace = "concat(INFO, ';')"
                     regex_replace = f"regexp_replace({regex_replace}, '{field_pattern}', '{field_renamed_pattern}')"
                     regex_replace_dict[regex_replace_key] = regex_replace
 
@@ -11747,7 +11747,7 @@ def rename_info_fields(
                 query = f"""
                     UPDATE {table}
                     SET
-                        INFO = {regex_replace}
+                        INFO = regexp_replace({regex_replace}, ';$', '')
                 """
                 log.debug(f"query={query}")
                 self.execute_query(query=query)

diff --git a/tests/data/example.annotation_names.vcf b/tests/data/example.annotation_names.vcf
@@ -56,8 +56,8 @@
 ##bcftools_viewVersion=1.15.1+htslib-1.15.1
 ##bcftools_viewCommand=view tests/data/example.vcf.gz; Date=Fri Mar 10 21:25:44 2023
 #CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	sample1	sample2	sample3	sample4
-chr1	28736	.	A	C	100	PASS	CLNSIG=pathogenic	GT:AD:DP:GQ	0/1:525,204:729:99	0/1:12659,4994:17664:99	1/1:12658,4995:17663:99	1/1:401,175:576:99
-chr1	35144	.	A	C	100	PASS	CLNSIG=non-pathogenic	GT:AD:DP:GQ	./.:.:.:.	0/1:12659,4994:17664:99	0/1:12658,4995:17663:99	0/1:401,175:576:99
+chr1	28736	.	A	C	100	PASS	SPiP_Alt=T;CLNSIG=pathogenic	GT:AD:DP:GQ	0/1:525,204:729:99	0/1:12659,4994:17664:99	1/1:12658,4995:17663:99	1/1:401,175:576:99
+chr1	35144	.	A	C	100	PASS	CLNSIG=non-pathogenic;SPiP_Alt=T	GT:AD:DP:GQ	./.:.:.:.	0/1:12659,4994:17664:99	0/1:12658,4995:17663:99	0/1:401,175:576:99
 chr1	69101	.	A	G	100	PASS	DP=50;CLNSIG=non-pathogenic;SIFT=D;SPiP_Alt	GT:AD:DP:GQ	0/1:525,204:729:99	./.:.:.:.	0/1:12658,4995:17663:99	0/1:401,175:576:99
 chr1	768251	.	A	G	100	PASS	CLNSIG=NP;PREFIXCLNSIG=NP;SPiP_Alt;CLNSIGSUFFIX=P	GT:AD:DP:GQ	0/1:525,204:729:99	./.:.:.:.	0/1:12658,4995:17663:99	0/1:401,175:576:99
 chr1	768252	.	A	G	100	PASS	PREFIXCLNSIG=NP	GT:AD:DP:GQ	0/1:525,204:729:99	./.:.:.:.	0/1:12658,4995:17663:99	0/1:401,175:576:99