Require metrics file in standalone pvacseq update_tiers step so that it can be updated with newly selected tiering parameters

Author: susannasiebert

pvactools/lib/update_tiers.py

@@ -6,6 +6,7 @@
 import tempfile
 import shutil
 import argparse
+import json
 
 from pvactools.lib.prediction_class import PredictionClass
 from pvactools.lib.run_utils import is_preferred_transcript, float_range, transcript_prioritization_strategy
@@ -33,6 +34,7 @@ def execute(self):
         output_df.to_csv(self.output_file, sep='\t', na_rep='NA', index=False, float_format='%.3f')
         shutil.copy(self.output_file.name, self.input_file)
         self.output_file.close()
+        self.update_metrics_file()
 
     @abstractmethod
     def get_tier(self, mutation):
@@ -42,6 +44,9 @@ def get_tier(self, mutation):
     def sort_table(self, output_lines):
         raise Exception("Must implement method in child class")
 
+    def update_metrics_file(self):
+        pass
+
     @classmethod
     def parser(cls, tool):
         parser = argparse.ArgumentParser(
@@ -53,6 +58,11 @@ def parser(cls, tool):
             'input_file',
             help="Input aggregated file with tiers to update. This file will be overwritten with the output."
         )
+        if tool in ['pvacseq']:
+            parser.add_argument(
+                'metrics_file',
+                help="metrics.json file corresponding to the input aggregated file. This file will be overwritten to update tiering parameters used by this command."
+            )
         if tool in ['pvacseq', 'pvacsplice']:
             parser.add_argument(
                 'vaf_clonal', type=float_range(0.0, 1.0),
@@ -167,6 +177,7 @@ def __init__(
             allele_specific_anchors=False,
             anchor_contribution_threshold=0.8,
             top_score_metric2='ic50',
+            metrics_file=None,
         ):
         self.input_file = input_file
         self.output_file = tempfile.NamedTemporaryFile()
@@ -180,6 +191,7 @@ def __init__(
         self.trna_vaf = trna_vaf
         self.transcript_prioritization_strategy = transcript_prioritization_strategy
         self.maximum_transcript_support_level = maximum_transcript_support_level
+        self.metrics_file=metrics_file
         if top_score_metric2 == "percentile":
             self.top_score_mode = "%ile MT"
         else:
@@ -341,6 +353,28 @@ def sort_table(self, output_lines):
 
         return df
 
+    def update_metrics_file(self):
+        if self.metrics_file is not None:
+            output_metrics_file = tempfile.NamedTemporaryFile()
+            with open(self.metrics_file, 'r') as input_fh, open(output_metrics_file.name, 'w') as output_fh:
+                metrics = json.loads(input_fh.read())
+                metrics['vaf_clonal'] = round(self.vaf_clonal, 3)
+                metrics['vaf_subclonal'] = round(self.vaf_clonal/2, 3)
+                metrics['binding_threshold'] = self.binding_threshold
+                metrics['trna_vaf'] = self.trna_vaf
+                metrics['trna_cov'] = self.trna_cov
+                metrics['allele_expr_threshold'] = self.allele_expr_threshold
+                metrics['transcript_prioritization_strategy'] = sorted(self.transcript_prioritization_strategy)
+                metrics['maximum_transcript_support_level'] = self.maximum_transcript_support_level
+                metrics['percentile_threshold'] = self.percentile_threshold
+                metrics['percentile_threshold_strategy'] = self.percentile_threshold_strategy
+                metrics['use_allele_specific_binding_thresholds'] = self.use_allele_specific_binding_thresholds
+                metrics['top_score_metric2'] = 'ic50' if self.top_score_mode == "IC50 MT" else 'percentile'
+                metrics['allele_specific_anchors'] = self.anchor_calculator.use_allele_specific_anchors
+                metrics['anchor_contribution_threshold'] = self.anchor_calculator.anchor_contribution_threshold
+                json.dump(metrics, output_fh, indent=2, separators=(',', ': '))
+            shutil.copy(output_metrics_file.name, self.metrics_file)
+
 class PvacfuseUpdateTiers(UpdateTiers, metaclass=ABCMeta):
     def __init__(
         self,

pvactools/tools/pvacseq/update_tiers.py

@@ -26,6 +26,7 @@ def main(args_input = sys.argv[1:]):
         allele_specific_anchors=args.allele_specific_anchors,
         anchor_contribution_threshold=args.anchor_contribution_threshold,
         top_score_metric2=args.top_score_metric2,
+        metrics_file=args.metrics_file,
     ).execute()
 
 if __name__ == "__main__":

tests/test_pvacseq_update_tiers.py

@@ -53,4 +53,5 @@ def test_compiles(self):
 
     def test_runs(self):
         input_file = os.path.join(self.test_data_directory, 'MHC_Class_I', 'Test.MHC_I.all_epitopes.aggregated.tsv')
-        self.assertFalse(update_tiers.main([input_file, "0.5"]))
+        input_metrics_file = os.path.join(self.test_data_directory, 'MHC_Class_I', 'Test.MHC_I.all_epitopes.aggregated.metrics.json')
+        self.assertFalse(update_tiers.main([input_file, input_metrics_file, "0.5"]))

tests/test_update_tiers.py

@@ -25,15 +25,24 @@ def test_update_tiers_pvacseq(self):
         input_file = os.path.join(self.test_data_dir, 'HCC1395.all_epitopes.aggregated.tsv')
         tmp_input_file = tempfile.NamedTemporaryFile()
         shutil.copy(input_file, tmp_input_file.name)
+        input_metrics_file = os.path.join(self.test_data_dir, 'HCC1395.all_epitopes.aggregated.metrics.json')
+        tmp_input_metrics_file = tempfile.NamedTemporaryFile()
+        shutil.copy(input_metrics_file, tmp_input_metrics_file.name)
         self.assertFalse(PvacseqUpdateTiers(
             tmp_input_file.name,
-            0.5
+            0.5,
+            metrics_file=tmp_input_metrics_file.name
         ).execute())
         self.assertTrue(cmp(
             tmp_input_file.name,
             os.path.join(self.test_data_dir, "HCC1395.all_epitopes.aggregated.out.tsv"),
         ))
+        self.assertTrue(cmp(
+            tmp_input_metrics_file.name,
+            os.path.join(self.test_data_dir, "HCC1395.all_epitopes.aggregated.metrics.out.json"),
+        ))
         tmp_input_file.close()
+        tmp_input_metrics_file.close()
 
     def test_update_tiers_pvacsplice(self):
         input_file = os.path.join(self.test_data_dir, 'HCC1395.pvacsplice.all_epitopes.aggregated.tsv')