updated tests and Filter class strategy param

ldhtnp · Feb 4, 2025 · 9392213 · 9392213
1 parent 1182f16
commit 9392213
Show file tree

Hide file tree

Showing 14 changed files with 21 additions and 14 deletions.
diff --git a/pvactools/lib/binding_filter.py b/pvactools/lib/binding_filter.py
@@ -7,7 +7,7 @@
 from pvactools.lib.run_utils import *
 
 class BindingFilter:
-    def __init__(self, input_file, output_file, binding_threshold, minimum_fold_change, top_score_metric, exclude_nas, allele_specific_cutoffs, percentile_threshold, percentile_threshold_strategy, file_type='pVACseq'):
+    def __init__(self, input_file, output_file, binding_threshold, minimum_fold_change, top_score_metric, exclude_nas, allele_specific_cutoffs, percentile_threshold, percentile_threshold_strategy='conservative', file_type='pVACseq'):
         self.input_file = input_file
         self.output_file = output_file
         self.binding_threshold = binding_threshold
@@ -49,8 +49,7 @@ def execute(self):
                 elif self.top_score_metric == 'lowest':
                     column = 'Corresponding Fold Change'
                 filter_criteria.append(FilterCriterion(column, '>=', self.minimum_fold_change, exclude_nas=self.exclude_nas))
-
-            Filter(self.input_file, self.output_file, filter_criteria, [], 0 if self.percentile_threshold_strategy == 'conservative' else 1).execute()
+            Filter(self.input_file, self.output_file, filter_criteria, [], "AND" if self.percentile_threshold_strategy == 'conservative' else "OR").execute()
 
     @classmethod
     def parser(cls, tool):

diff --git a/pvactools/lib/filter.py b/pvactools/lib/filter.py
@@ -5,7 +5,7 @@
 pd.options.mode.chained_assignment = None
 
 class Filter:
-    def __init__(self, input_file, output_file, filter_criteria, int_filter_columns=[], filter_strategy=0):
+    def __init__(self, input_file, output_file, filter_criteria, int_filter_columns=[], filter_strategy="AND"):
         self.input_file = input_file
         self.output_file = output_file
         self.filter_criteria = filter_criteria
@@ -22,18 +22,17 @@ def execute(self):
 
                 process_criterion = lambda criterion: (
                     (line[criterion.column] == 'NA' and criterion.exclude_nas) or
-                    (line[criterion.column] != 'NA' and criterion.skip_value is not None and line[criterion.column] == criterion.skip_value) or
-                    (line[criterion.column] != 'NA' and criterion.skip_value is None and not eval("{} {} {}".format(
+                    (line[criterion.column] != 'NA' and criterion.skip_value != line[criterion.column] and not eval("{} {} {}".format(
                         line[criterion.column] if line[criterion.column] != 'inf' else sys.maxsize,
                         criterion.operator,
                         criterion.threshold
                     )))
                 )
 
-                if self.filter_strategy:
-                    to_filter = all(process_criterion(criterion) for criterion in self.filter_criteria)
-                else:
+                if self.filter_strategy == "AND":
                     to_filter = any(process_criterion(criterion) for criterion in self.filter_criteria)
+                else:
+                    to_filter = all(process_criterion(criterion) for criterion in self.filter_criteria)
 
                 if not to_filter:
                     writer.writerow(line)

diff --git a/pvactools/tools/pvacbind/binding_filter.py b/pvactools/tools/pvacbind/binding_filter.py
@@ -9,7 +9,7 @@ def main(args_input = sys.argv[1:]):
     parser = define_parser()
     args = parser.parse_args(args_input)
 
-    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, 'pVACbind').execute()
+    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, args.percentile_threshold_strategy, 'pVACbind').execute()
 
 if __name__ == "__main__":
     main()
diff --git a/pvactools/tools/pvacfuse/binding_filter.py b/pvactools/tools/pvacfuse/binding_filter.py
@@ -9,7 +9,7 @@ def main(args_input = sys.argv[1:]):
     parser = define_parser()
     args = parser.parse_args(args_input)
 
-    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, 'pVACfuse').execute()
+    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, args.percentile_threshold_strategy, 'pVACfuse').execute()
 
 if __name__ == "__main__":
     main()
diff --git a/pvactools/tools/pvacseq/binding_filter.py b/pvactools/tools/pvacseq/binding_filter.py
@@ -9,7 +9,7 @@ def main(args_input = sys.argv[1:]):
     parser = define_parser()
     args = parser.parse_args(args_input)
 
-    BindingFilter(args.input_file, args.output_file, args.binding_threshold, args.minimum_fold_change, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold).execute()
+    BindingFilter(args.input_file, args.output_file, args.binding_threshold, args.minimum_fold_change, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, args.percentile_threshold_strategy).execute()
 
 if __name__ == "__main__":
     main()
diff --git a/pvactools/tools/pvacsplice/binding_filter.py b/pvactools/tools/pvacsplice/binding_filter.py
@@ -9,7 +9,7 @@ def main(args_input = sys.argv[1:]):
     parser = define_parser()
     args = parser.parse_args(args_input)
 
-    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, file_type='pVACsplice').execute()
+    BindingFilter(args.input_file, args.output_file, args.binding_threshold, None, args.top_score_metric, args.exclude_NAs, args.allele_specific_binding_thresholds, args.percentile_threshold, args.percentile_threshold_strategy, 'pVACsplice').execute()
 
 if __name__ == "__main__":
     main()
diff --git a/tests/test_data/aggregate_all_epitopes/HCC1395.output.metrics.json b/tests/test_data/aggregate_all_epitopes/HCC1395.output.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Median",
   "wt_top_score_metric": "Median",

diff --git a/...s/test_data/aggregate_all_epitopes/output.allele_specific_binding_thresholds.metrics.json b/...s/test_data/aggregate_all_epitopes/output.allele_specific_binding_thresholds.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": true,
   "mt_top_score_metric": "Median",
   "wt_top_score_metric": "Median",

diff --git a/tests/test_data/aggregate_all_epitopes/output.lowest_top_score_metric.metrics.json b/tests/test_data/aggregate_all_epitopes/output.lowest_top_score_metric.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Best",
   "wt_top_score_metric": "Corresponding",

diff --git a/tests/test_data/aggregate_all_epitopes/output.metrics.json b/tests/test_data/aggregate_all_epitopes/output.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Median",
   "wt_top_score_metric": "Median",

diff --git a/tests/test_data/aggregate_all_epitopes/output.na_mutation_position.metrics.json b/tests/test_data/aggregate_all_epitopes/output.na_mutation_position.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Median",
   "wt_top_score_metric": "Median",

diff --git a/tests/test_data/pvacseq/MHC_Class_I/Test.all_epitopes.aggregated.metrics.json b/tests/test_data/pvacseq/MHC_Class_I/Test.all_epitopes.aggregated.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Best",
   "wt_top_score_metric": "Corresponding",

diff --git a/tests/test_data/pvacseq/MHC_Class_II/Test.all_epitopes.aggregated.metrics.json b/tests/test_data/pvacseq/MHC_Class_II/Test.all_epitopes.aggregated.metrics.json
@@ -10,6 +10,7 @@
   "allele_expr_threshold": 2.5,
   "maximum_transcript_support_level": 1,
   "percentile_threshold": null,
+  "percentile_threshold_strategy": "conservative",
   "use_allele_specific_binding_thresholds": false,
   "mt_top_score_metric": "Best",
   "wt_top_score_metric": "Corresponding",

diff --git a/tests/test_filter.py b/tests/test_filter.py
@@ -203,6 +203,8 @@ def test_conservative(self):
                 "16000",
                 exclude_nas=False
             )],
+            [],
+            "AND"
         ).execute())
         self.assertTrue(cmp(
             output_file.name,
@@ -230,7 +232,7 @@ def test_exploratory(self):
                 exclude_nas=False
             )],
             [],
-            1
+            "OR"
         ).execute())
         self.assertTrue(cmp(
             output_file.name,