linting, standardizing

Sage-Bionetworks · Nov 7, 2023 · 780ee66 · 780ee66
1 parent 4fea695
commit 780ee66
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 21 deletions.
diff --git a/genie_registry/maf.py b/genie_registry/maf.py
@@ -64,7 +64,7 @@ class maf(FileTypeFormat):
     _process_kwargs = []
     _allele_cols = ["REFERENCE_ALLELE", "TUMOR_SEQ_ALLELE1", "TUMOR_SEQ_ALLELE2"]
     _allowed_comb_alleles = ["A", "T", "C", "G", "N"]
-    _allowed_ind_alleles = ['-']
+    _allowed_ind_alleles = ["-"]
 
     def _validateFilename(self, filePath):
         """
@@ -298,7 +298,7 @@ def _validate(self, mutationDF):
                     allowed_comb_alleles=self._allowed_comb_alleles,
                     allowed_ind_alleles=self._allowed_ind_alleles,
                     ignore_case=True,
-                    allow_na=False
+                    allow_na=False,
                 )
                 errors, warnings = validate.get_allele_validation_message(
                     invalid_indices,

diff --git a/genie_registry/vcf.py b/genie_registry/vcf.py
@@ -18,7 +18,7 @@ class vcf(FileTypeFormat):
     _fileType = "vcf"
 
     _process_kwargs = []
-    _allele_col = "REF"
+    _allele_cols = ["REF"]
     _allowed_comb_alleles = ["A", "T", "C", "G", "N"]
     _allowed_ind_alleles = []
 
@@ -140,24 +140,25 @@ def _validate(self, vcfdf):
         total_error += error
         warning += warn
 
-        if process_functions.checkColExist(vcfdf, self._allele_col):
-            invalid_indices = validate.get_invalid_allele_rows(
-                vcfdf,
-                input_col=self._allele_col,
-                allowed_comb_alleles=self._allowed_comb_alleles,
-                allowed_ind_alleles=self._allowed_ind_alleles,
-                ignore_case=True,
-                allow_na=False
+        for allele_col in self._allele_cols:
+            if process_functions.checkColExist(vcfdf, allele_col):
+                invalid_indices = validate.get_invalid_allele_rows(
+                    vcfdf,
+                    input_col=allele_col,
+                    allowed_comb_alleles=self._allowed_comb_alleles,
+                    allowed_ind_alleles=self._allowed_ind_alleles,
+                    ignore_case=True,
+                    allow_na=False,
                 )
-            errors, warnings = validate.get_allele_validation_message(
-                invalid_indices,
-                invalid_col=self._allele_col,
-                allowed_comb_alleles=self._allowed_comb_alleles,
-                allowed_ind_alleles=self._allowed_ind_alleles,
-                fileformat=self._fileType,
-            )
-            total_error += errors
-            warning += warnings
+                errors, warnings = validate.get_allele_validation_message(
+                    invalid_indices,
+                    invalid_col=allele_col,
+                    allowed_comb_alleles=self._allowed_comb_alleles,
+                    allowed_ind_alleles=self._allowed_ind_alleles,
+                    fileformat=self._fileType,
+                )
+                total_error += errors
+                warning += warnings
 
         # No white spaces
         white_space = vcfdf.apply(lambda x: contains_whitespace(x), axis=1)

diff --git a/tests/test_maf.py b/tests/test_maf.py
@@ -19,7 +19,13 @@ def valid_maf_df():
         dict(
             CHROMOSOME=[1, 2, 3, 4, 5],
             START_POSITION=[1, 2, 3, 4, 2],
-            REFERENCE_ALLELE=["A", "A", "A", "A", "A"],
+            REFERENCE_ALLELE=[
+                "C",
+                "G",
+                "NA",
+                "-",
+                "TAAAGATCGTACAGAA",
+            ],
             TUMOR_SAMPLE_BARCODE=[
                 "GENIE-SAGE-ID1-1",
                 "GENIE-SAGE-ID1-1",