Merge pull request #323 from bioinfo-chru-strasbourg/fix_rename_info_…

…tags fix info/tags rename #307, docs #4
bioinfo-chru-strasbourg · Dec 9, 2024 · bb96703 · bb96703
2 parents c159f13 + 81498ee
commit bb96703
Show file tree

Hide file tree

Showing 16 changed files with 247 additions and 221 deletions.
diff --git a/README.pdf b/README.pdf
diff --git a/RELEASE_NOTES.pdf b/RELEASE_NOTES.pdf
diff --git a/docs/docs.pdf b/docs/docs.pdf
diff --git a/docs/help.configuration.calculation.pdf b/docs/help.configuration.calculation.pdf
diff --git a/docs/help.configuration.pdf b/docs/help.configuration.pdf
diff --git a/docs/help.configuration.prioritization.pdf b/docs/help.configuration.prioritization.pdf
diff --git a/docs/help.parameters.databases.pdf b/docs/help.parameters.databases.pdf
diff --git a/docs/help.parameters.pdf b/docs/help.parameters.pdf
diff --git a/docs/help.pdf b/docs/help.pdf
diff --git a/docs/pdoc/howard/objects/variants.html b/docs/pdoc/howard/objects/variants.html
diff --git a/docs/tips.pdf b/docs/tips.pdf
diff --git a/docs/user_guide.pdf b/docs/user_guide.pdf
diff --git a/howard/objects/variants.py b/howard/objects/variants.py
@@ -11713,9 +11713,9 @@ 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 = ''
 
@@ -11736,6 +11736,11 @@ def rename_info_fields(
                     else:
                         log.info(f"Rename or remove fields - field '{field_to_rename}' removed")
 
+                else:
+
+                    log.warning(f"Rename or remove fields - field '{field_to_rename}' not in header")
+
+
             # Rename INFO
             for regex_replace_key, regex_replace  in regex_replace_dict.items():
                 log.info(f"Rename or remove fields - Process [{regex_replace_key+1}/{len(regex_replace_dict)}]...")

diff --git a/plugins/README.pdf b/plugins/README.pdf
diff --git a/tests/data/example.annotation_names.vcf b/tests/data/example.annotation_names.vcf
@@ -24,6 +24,7 @@
 ##INFO=<ID=CLNSIG,Number=1,Type=String,Description="CLNSIG">
 ##INFO=<ID=PREFIXCLNSIG,Number=1,Type=String,Description="CLNSIG">
 ##INFO=<ID=CLNSIGSUFFIX,Number=1,Type=String,Description="CLNSIG">
+##INFO=<ID=SPiP_Alt,Number=0,Type=Flag,Description="SPiP">
 ##contig=<ID=chr1,length=249250621,assembly=hg19>
 ##contig=<ID=chr7,length=159138663,assembly=hg19>
 ##contig=<ID=1,length=249250621,assembly=hg19>
@@ -57,8 +58,8 @@
 #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	69101	.	A	G	100	PASS	DP=50;CLNSIG=non-pathogenic;SIFT=D	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;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	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
-chr1	768253	.	A	G	100	PASS	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	768253	.	A	G	100	PASS	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
 chr7	55249063	rs1050171	G	A	5777	PASS	CLNSIG=NP;DP=125;CLNSIG=NP	GT:AD:DP:GQ	0/1:525,204:729:99	0/1:12659,4994:17664:99	./.:.:.:.	0/1:401,175:576:99
diff --git a/tests/test_objects_variants.py b/tests/test_objects_variants.py
@@ -2196,13 +2196,16 @@ def test_rename_fields():
             "field_not_in_header": "field_not_in_header_renamed",
             "": "",
             "SIFT": None,
+            "SPiP_Alt": "SPiP_alternative",
+            "SPiP_alternative": None,
         }
 
         # Rename fields
         fields_renamed = variants.rename_info_fields(fields_to_rename=fields_to_rename)
-        assert fields_renamed == {'CLNSIG': 'CLNSIG_renamed', 'PREFIXCLNSIG': 'PREFIXCLNSIG_renamed', 'DP': 'depth', 'SIFT': None}
+        assert fields_renamed == {'CLNSIG': 'CLNSIG_renamed', 'PREFIXCLNSIG': 'PREFIXCLNSIG_renamed', 'DP': 'depth', 'SIFT': None, 'SPiP_Alt': 'SPiP_alternative', 'SPiP_alternative': None}
         assert len(variants.get_query_to_df("SELECT INFO FROM variants WHERE INFO LIKE '%SIFT%'")) == 0
         assert len(variants.get_query_to_df("SELECT INFO FROM variants WHERE INFO LIKE '%None=%'")) == 0
+        assert len(variants.get_query_to_df("SELECT INFO FROM variants WHERE INFO LIKE '%SPiP%'")) == 0
 
         # Check if VCF is in correct format with pyVCF
         remove_if_exists([output_vcf])
@@ -2235,6 +2238,8 @@ def test_rename_fields_to_param_and_export():
                     "field_not_in_header": "field_not_in_header_renamed",
                     "": "",
                     "SIFT": None,
+                    "SPiP_Alt": "SPiP_alternative",
+                    "SPiP_alternative": None,
                 }
             }
         }