Merge pull request #515 from GavinHuttley/develop

improve robustness of ArrayAlignment methods

src/cogent3/core/alignment.py

@@ -3039,7 +3039,9 @@ def counts_per_seq(
         motifs = set()
         for name in self.names:
             if is_array:
-                seq = self.named_seqs[name]
+                seq = self.moltype.make_array_seq(
+                    self.array_seqs[self.names.index(name)]
+                )
             else:
                 seq = self.get_gapped_seq(name)
             c = seq.counts(

src/cogent3/core/moltype.py

@@ -714,7 +714,7 @@ def __init__(
             self.array_type = None
 
         # set modeling sequence
-        self.make_array_seq = array_seq_constructor
+        self._make_array_seq = array_seq_constructor
 
         self._colors = colors or defaultdict(_DefaultValue("black"))
 
@@ -762,6 +762,29 @@ def make_seq(self, seq, name=None, **kwargs):
         """Returns sequence of correct type."""
         return self._make_seq(seq, name, **kwargs)
 
+    def make_array_seq(self, seq, name=None, **kwargs):
+        """
+        creates an array sequence
+
+        Parameters
+        ----------
+        seq
+            characters or array
+        name : str
+        kwargs
+            keyword arguments for the ArraySequence constructor.
+
+        Returns
+        -------
+        ArraySequence
+        """
+        alphabet = kwargs.pop("alphabet", None)
+        if alphabet is None and hasattr(self, "alphabets"):
+            alphabet = self.alphabets.degen_gapped
+        elif alphabet is None:
+            alphabet = self.alphabet
+        return self._make_array_seq(seq, alphabet=alphabet, name=name, **kwargs)
+
     def verify_sequence(self, seq, gaps_allowed=True, wildcards_allowed=True):
         """Checks whether sequence is valid on the default alphabet.
 

tests/test_core/test_sequence.py

@@ -12,8 +12,17 @@
 from numpy.testing import assert_allclose
 
 from cogent3.core.annotation import Feature, SimpleVariable, Variable
-from cogent3.core.moltype import ASCII, BYTES, DNA, PROTEIN, RNA, AlphabetError
+from cogent3.core.moltype import (
+    ASCII,
+    BYTES,
+    DNA,
+    PROTEIN,
+    RNA,
+    AlphabetError,
+    get_moltype,
+)
 from cogent3.core.sequence import (
+    ABSequence,
     ArrayCodonSequence,
     ArrayDnaCodonSequence,
     ArrayDnaSequence,
@@ -1110,6 +1119,7 @@ class ModelSequenceTests(SequenceTests):
     RNA = ArrayRnaSequence
     DNA = ArrayDnaSequence
     PROT = ArrayProteinSequence
+    AB = ABSequence
 
     def test_distance_indices(self):
         """ArraySequence distance should work with function of indices"""
@@ -1163,6 +1173,15 @@ def test_gap_indices(self):
         v = r.gap_indices()
         self.assertEqual(v, array([0, 1]))
 
+    def test_count_ab(self):
+        """abseq array seq should count characters"""
+        AB = get_moltype("ab")
+        seq = AB.make_array_seq("aaba-", alphabet=AB.alphabet.with_gap_motif())
+        c = seq.counts()
+        self.assertEqual(c.to_dict(), {"a": 3, "b": 1})
+        c = seq.counts(allow_gap=True)
+        self.assertEqual(c.to_dict(), {"a": 3, "b": 1, "-": 1})
+
 
 # run if called from command-line
 if __name__ == "__main__":