Warn on writing INSDC-noncompliant feature and qualifier keys (biopyt…

…hon#4703)

See https://www.insdc.org/submitting-standards/feature-table/#3.1 for
what can and cannot be in an INSDC feature or qualifier key.

Bio/SeqIO/InsdcIO.py

@@ -33,6 +33,7 @@
 import warnings
 
 from datetime import datetime
+from string import ascii_letters, digits
 
 from Bio import BiopythonWarning
 from Bio import SeqFeature
@@ -46,6 +47,9 @@
 from .Interfaces import SequenceIterator
 from .Interfaces import SequenceWriter
 
+# Set containing all characters allowed in feature qualifier keys. See
+# https://www.insdc.org/submitting-standards/feature-table/#3.1
+_allowed_table_component_name_chars = set(ascii_letters + digits + "_-'*")
 
 # NOTE
 # ====
@@ -376,6 +380,19 @@ class _InsdcWriter(SequenceWriter):
     )
 
     def _write_feature_qualifier(self, key, value=None, quote=None):
+        if not _allowed_table_component_name_chars.issuperset(key):
+            warnings.warn(
+                f"Feature qualifier key '{key}' contains characters not"
+                " allowed by standard.",
+                BiopythonWarning,
+            )
+        if len(key) > 20:
+            warnings.warn(
+                f"Feature qualifier key '{key}' is longer than maximum length"
+                " specified by standard (20 characters).",
+                BiopythonWarning,
+            )
+
         if value is None:
             # Value-less entry like /pseudo
             self.handle.write(f"{self.QUALIFIER_INDENT_STR}/{key}\n")
@@ -439,8 +456,22 @@ def _wrap_location(self, location):
     def _write_feature(self, feature, record_length):
         """Write a single SeqFeature object to features table (PRIVATE)."""
         assert feature.type, feature
-        location = _insdc_location_string(feature.location, record_length)
+
         f_type = feature.type.replace(" ", "_")
+        if not _allowed_table_component_name_chars.issuperset(f_type):
+            warnings.warn(
+                f"Feature key '{f_type}' contains characters not allowed by"
+                " standard.",
+                BiopythonWarning,
+            )
+        if len(f_type) > 15:
+            warnings.warn(
+                f"Feature key '{f_type}' is longer than maximum length"
+                " specified by standard (15 characters).",
+                BiopythonWarning,
+            )
+
+        location = _insdc_location_string(feature.location, record_length)
         line = (
             (self.QUALIFIER_INDENT_TMP % f_type)[: self.QUALIFIER_INDENT]
             + self._wrap_location(location)

Tests/test_SeqIO_Insdc.py

@@ -9,7 +9,7 @@
 
 from io import StringIO
 
-from Bio import BiopythonParserWarning
+from Bio import BiopythonWarning, BiopythonParserWarning
 from Bio import SeqIO
 from Bio.Seq import Seq
 from Bio.SeqFeature import SimpleLocation
@@ -75,6 +75,28 @@ def test_writing_empty_qualifiers(self):
         self.assertIn(" /one=1\n", gbk)
         self.assertIn(' /text="blah"\n', gbk)
 
+    def test_warn_on_writing_nonstandard_feature_key(self):
+        f = SeqFeature(
+            SimpleLocation(5, 20, strand=+1),
+            type="a" * 16,
+            qualifiers={"empty": None, "zero": 0, "one": 1, "text": "blah"},
+        )
+        record = SeqRecord(Seq("A" * 100), "dummy", features=[f])
+        record.annotations["molecule_type"] = "DNA"
+        with self.assertWarns(BiopythonWarning):
+            record.format("gb")
+
+    def test_warn_on_writing_nonstandard_qualifier_key(self):
+        f = SeqFeature(
+            SimpleLocation(5, 20, strand=+1),
+            type="region",
+            qualifiers={"a" * 21: "test"},
+        )
+        record = SeqRecord(Seq("A" * 100), "dummy", features=[f])
+        record.annotations["molecule_type"] = "DNA"
+        with self.assertWarns(BiopythonWarning):
+            record.format("gb")
+
 
 class TestEmblRewrite(SeqRecordTestBaseClass):
     def check_rewrite(self, filename):