Add skipping internal_stop to AFP

pha4ge · Oct 18, 2023 · 7c4d710 · 7c4d710
1 parent 48477e4
commit 7c4d710
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 0 deletions.
diff --git a/hAMRonization/AmrFinderPlusIO.py b/hAMRonization/AmrFinderPlusIO.py
@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 
 import csv
+import warnings
 import re
 from .Interfaces import hAMRonizedResultIterator
 from hAMRonization.constants import (
@@ -93,13 +94,21 @@ def parse(self, handle):
         """
         Read each and return it
         """
+        skipped_truncated = 0
         reader = csv.DictReader(handle, delimiter="\t")
         for result in reader:
             # replace NA value with None for consitency
             for field, value in result.items():
                 if value == "NA":
                     result[field] = None
 
+            # AFP reports partial hits so to avoid misleadingly listing these
+            # as present skip results with INTERNAL_STOP
+            # recommended by developers
+            if "INTERNAL_STOP" in result['Method']:
+                skipped_truncated += 1
+                continue
+
             # "POINT" indicates mutational resistance
             # amrfinderplus has no special fields but the mutation itself is
             # appended to the symbol name so we want to split this
@@ -121,3 +130,8 @@ def parse(self, handle):
                     result["genetic_variation_type"] = NUCLEOTIDE_VARIANT
 
             yield self.hAMRonize(result, self.metadata)
+
+        if skipped_truncated > 0:
+            warnings.warn(f"Skipping {skipped_truncated} records with INTERNAL_STOP "
+                          f"from {self.metadata['input_file_name']}")
+
diff --git a/test/data/raw_outputs/amrfinderplus/report_protein.tsv b/test/data/raw_outputs/amrfinderplus/report_protein.tsv
@@ -11,3 +11,4 @@ emrD3-suppressed-in-vibrio	contig13	1	1137	+	emrD3	multidrug efflux MFS transpor
 pmrB_C84R	contig14	1093	2181	+	pmrB_C84R	Escherichia colistin resistant PmrB	core	AMR	POINT	COLISTIN	COLISTIN	POINTP	363	363	100.00	99.72	363	WP_001300761.1	two-component system sensor histidine kinase PmrB	NA	NA
 nfsA_R15C_K141STOP	contig16	1	423	+	nfsA_R15C	Escherichia nitrofurantoin resistant NfsA	core	AMR	POINT	NITROFURAN	NITROFURANTOIN	POINTP	140	240	58.33	99.29	140	WP_089631889.1	nitroreductase NfsA	NA	NA
 nimIJ_hmm	contigX	1	501	+	nimIJ	NimIJ family nitroimidazole resistance protein	core	AMR	AMR	NITROIMIDAZOLE	NITROIMIDAZOLE	HMM	166	165	98.18	76.54	162	WP_005812825.1	NimIJ family nitroimidazole resistance protein	NF000262.1	NimIJ family nitroimidazole resistance protein
+nimIJ_skip	contigX	1	501	+	nimIJ	NimIJ family nitroimidazole resistance protein	core	AMR	AMR	NITROIMIDAZOLE	NITROIMIDAZOLE	INTERNAL_STOP	166	165	98.18	76.54	162	WP_005812825.1	NimIJ family nitroimidazole resistance protein	NF000262.1	NimIJ family nitroimidazole resistance protein