feat (parseFusionCatcher): added --skip-failed

Author: zhuchcn

CHANGELOG.md

@@ -12,7 +12,7 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ## [1.4.6-rc3] - 2025-03-10
 
-- Added --skip-failed flag to callVariant, parseArriba, parserSTARFusion.
+- Added --skip-failed flag to callVariant, parseArriba, parserSTARFusion, parseFusionCatcher.
 
 ## [1.4.6-rc2] - 2025-03-03
 

moPepGen/cli/parse_arriba.py

@@ -146,8 +146,8 @@ def parse_arriba(args:argparse.Namespace) -> None:
                 if args.skip_failed:
                     tally.skipped.invalid_position += 1
                     tally.skipped.total += 1
-                else:
-                    raise
+                    continue
+                raise
 
     logger.info('Arriba output %s loaded.', fusion)
 

moPepGen/cli/parse_fusion_catcher.py

@@ -2,13 +2,18 @@
 [FusionCatcher](https://github.com/ndaniel/fusioncatcher) and save as a
 GVF file. The GVF file can be later used to call variant peptides using
 [callVariant](call-variant.md)."""
-from typing import List
+from __future__ import annotations
+from typing import TYPE_CHECKING
 from pathlib import Path
 import argparse
 from moPepGen import get_logger, seqvar, parser, err
 from moPepGen.cli import common
 
 
+if TYPE_CHECKING:
+    from typing import List
+    from logging import Logger
+
 INPUT_FILE_FORMATS = ['.tsv', '.txt']
 OUTPUT_FILE_FORMATS = ['.gvf']
 
@@ -42,16 +47,48 @@ def add_subparser_parse_fusion_catcher(subparsers:argparse._SubParsersAction):
         default=5,
         metavar='<number>'
     )
+    common.add_args_skip_failed(p)
     common.add_args_source(p)
     common.add_args_reference(p, proteome=False)
     common.add_args_debug_level(p)
     p.set_defaults(func=parse_fusion_catcher)
     common.print_help_if_missing_args(p)
     return p
 
+class TallyTable():
+    """ Tally table """
+    def __init__(self, logger:Logger):
+        """ Constructor """
+        self.total:int = 0
+        self.succeed:int = 0
+        self.skipped:TallyTableSkipped = TallyTableSkipped()
+        self.logger = logger
+
+    def log(self):
+        """ Show tally results """
+        self.logger.info("Summary:")
+        self.logger.info("Totally records read: %i", self.total)
+        self.logger.info("Records successfully processed: %i", self.succeed)
+        self.logger.info("Records skipped: %i", self.skipped.total)
+        if self.skipped.total > 0:
+            self.logger.info("Out of those skipped,")
+            self.logger.info("    Invalid gene ID: %i", self.skipped.invalid_gene_id)
+            self.logger.info("    Invalid position: %i", self.skipped.invalid_position)
+            self.logger.info("    Insufficient evidence: %i", self.skipped.insufficient_evidence)
+
+class TallyTableSkipped():
+    """ Tally table for failed ones """
+    def __init__(self):
+        """ constructor """
+        self.invalid_gene_id:int = 0
+        self.invalid_position:int = 0
+        self.insufficient_evidence:int = 0
+        self.total:int = 0
+
 def parse_fusion_catcher(args:argparse.Namespace) -> None:
     """ Parse FusionCatcher output and save it in GVF format. """
     logger = get_logger()
+    tally = TallyTable(logger)
     # unpack args
     fusion = args.input_path
     output_path:Path = args.output_path
@@ -69,15 +106,26 @@ def parse_fusion_catcher(args:argparse.Namespace) -> None:
     variants:List[seqvar.VariantRecord] = []
 
     for record in parser.FusionCatcherParser.parse(fusion):
-        if record.counts_of_common_mapping_reads > args.max_common_mapping:
-            continue
-        if record.spanning_unique_reads < args.min_spanning_unique:
+        tally.total += 1
+        if record.counts_of_common_mapping_reads > args.max_common_mapping \
+                or record.spanning_unique_reads < args.min_spanning_unique:
+            tally.skipped.insufficient_evidence += 1
+            tally.skipped.total += 1
             continue
         try:
             var_records = record.convert_to_variant_records(anno, genome)
+            variants.extend(var_records)
+            tally.succeed += 1
         except err.GeneNotFoundError:
+            tally.skipped.invalid_gene_id += 1
+            tally.skipped.total += 1
             continue
-        variants.extend(var_records)
+        except:
+            if args.skip_failed:
+                tally.skipped.total += 1
+                tally.skipped.invalid_position += 1
+                continue
+            raise
 
     logger.info('FusionCatcher output %s loaded.', fusion)
 
@@ -95,3 +143,5 @@ def parse_fusion_catcher(args:argparse.Namespace) -> None:
     seqvar.io.write(variants, output_path, metadata)
 
     logger.info("Variants written to disk.")
+
+    tally.log()

moPepGen/cli/parse_star_fusion.py

@@ -103,20 +103,22 @@ def parse_star_fusion(args:argparse.Namespace) -> None:
         tally.total += 1
         if record.est_j < args.min_est_j:
             tally.skipped.insufficient_evidence += 1
+            tally.skipped.total += 1
             continue
         try:
             var_records = record.convert_to_variant_records(anno, genome)
             variants.extend(var_records)
             tally.succeed += 1
         except err.GeneNotFoundError:
             tally.skipped.invalid_gene_id += 1
+            tally.skipped.total += 1
             continue
         except:
             if args.skip_failed:
                 tally.skipped.invalid_position += 1
+                tally.skipped.total += 1
                 continue
-            else:
-                raise
+            raise
 
     logger.info('STAR-Fusion output %s loaded.', fusion)
 

test/integration/test_parse_fusion_catcher.py

@@ -3,6 +3,7 @@
 from pathlib import Path
 import subprocess as sp
 import sys
+from unittest.mock import Mock
 from test.unit import load_references
 from test.integration import TestCaseIntegration
 from moPepGen import cli, parser
@@ -54,11 +55,9 @@ def test_fusioncatcher_parser(self):
                 right_seq = gene2_seq.seq[_start2:_end2]
                 self.assertEqual(str(right_seq), fusion_seq[1])
 
-    def test_parse_fusion_catcher(self):
-        """ Test parseFusionCatcher """
+    def create_base_args(self) -> argparse.Namespace:
         args = argparse.Namespace()
         args.command = 'parseFusionCatcher'
-        args.input_path = self.data_dir/'fusion/fusion_catcher.txt'
         args.source = 'Fusion'
         args.index_dir = None
         args.genome_fasta = self.data_dir/'genome.fasta'
@@ -68,9 +67,31 @@ def test_parse_fusion_catcher(self):
         args.output_path = self.work_dir/'fusion_catcher.gvf'
         args.max_common_mapping = 0
         args.min_spanning_unique = 5
+        args.skip_failed = False
         args.quiet = True
+        return args
+
+    def test_parse_fusion_catcher(self):
+        """ Test parseFusionCatcher """
+        args = self.create_base_args()
+        args.input_path = self.data_dir/'fusion/fusion_catcher.txt'
         cli.parse_fusion_catcher(args)
         files = {str(file.name) for file in self.work_dir.glob('*')}
         expected = {'fusion_catcher.gvf'}
         self.assertEqual(files, expected)
         self.assert_gvf_order(args.output_path, args.annotation_gtf)
+
+    def test_parse_fusion_catcher_skip_failed(self):
+        """ Test parseFusionCatcher with --skip-failed """
+        from moPepGen import parser
+        parser.FusionCatcherParser.FusionCatcherRecord.convert_to_variant_records = Mock(
+            side_effect=ValueError()
+        )
+
+        args = self.create_base_args()
+        args.input_path = self.data_dir/'fusion/fusion_catcher.txt'
+        with self.assertRaises(ValueError):
+            cli.parse_fusion_catcher(args)
+
+        args.skip_failed = True
+        cli.parse_fusion_catcher(args)