Merge branch 'dev' into main

polymera/__init__.py

@@ -1,4 +1,4 @@
-from .polymera import Polymer, Alphabet, Sequence, Segment
+from .polymera import Polymer, Alphabet, Sequence, Segment, hamming
 from .bio import dna
 
 from .version import __version__

polymera/polymera.py

@@ -323,3 +323,59 @@ def __init__(self, choices):
         self.choices = choices
         if len(set(choices)) != len(choices):
             raise DuplicateChoiceError("Choices of a segment must be unique.")
+
+
+class UnequalLengthError(ValueError):
+    pass
+
+
+def hamming(seq1, seq2, comparison="options"):
+    if seq1.get_length() != seq2.get_length():
+        raise UnequalLengthError("The two sequences must be the same length!")
+
+    if comparison == "options":
+        distance = hamming_options(seq1, seq2)
+    elif comparison == "uncertainty":
+        distance = 0
+    else:
+        raise ValueError("Parameter comparison must be 'options' or 'uncertainty'!")
+
+    return distance
+
+
+def break_segment(segment):
+    """Return a list of n 1-length Segments from a Segment of length n."""
+    new_segments = []
+    for i in range(0, len(segment.choices[0])):
+        new_choices = []
+        for choice in segment.choices:
+            new_choices += choice[i]
+            new_choices = list(set(new_choices))  # remove duplicate letters
+        new_segment = Segment(choices=new_choices)
+        new_segments += [new_segment]
+
+    return new_segments
+
+
+def convert_to_nosegment(seq):
+    """Convert Sequence segments to length 1 segments."""
+    new_segments = []
+    for segment in seq.segments:
+        new_subsegments = break_segment(segment)
+        new_segments += new_subsegments
+
+    sequence = Sequence(segments=new_segments)
+    return sequence
+
+
+def hamming_options(seq1, seq2):
+    sequence1 = convert_to_nosegment(seq1)
+    sequence2 = convert_to_nosegment(seq2)
+
+    distance = 0
+    for i, segment1 in enumerate(sequence1.segments):
+        segment2 = sequence2.segments[i]
+        if set(segment1.choices) & set(segment2.choices) == set():
+            distance += 1
+
+    return distance

tests/test_polymera.py

@@ -1,5 +1,6 @@
 import pytest
 import polymera
+from polymera.polymera import break_segment, convert_to_nosegment, hamming_options
 
 
 def test_segment():
@@ -62,3 +63,50 @@ def test_polymer():
         polymer.get_information_content(method="wrong parameter")
     assert polymer.get_information_content(method="option") == 2
     assert polymer.get_information_content(method="uncertainty") == 1
+
+
+def test_break_segment():
+    segment = polymera.Segment(["AAT", "AAC"])
+    new_segments = break_segment(segment)
+    assert len(new_segments) == 3
+    assert len(new_segments[2].choices) == 2
+
+
+def test_convert_to_nosegment():
+    seq1 = polymera.Sequence()
+    seq1.add_sequence_from_string("T,C,G|CCC")
+    new_seq = convert_to_nosegment(seq1)
+    assert len(new_seq.segments) == 4
+
+
+def test_hamming_options():
+    seq1 = polymera.Sequence()
+    seq1.add_sequence_from_string("T,C,G|CCC")
+
+    seq2 = polymera.Sequence()
+    seq2.add_sequence_from_string("A|GGG")
+
+    assert hamming_options(seq1, seq2) == 4
+
+
+def test_hamming():
+    seq1 = polymera.Sequence()
+    seq1.add_sequence_from_string("T,C,G|CCG")
+
+    seq2 = polymera.Sequence()
+    seq2.add_sequence_from_string("A|GGG")
+
+    # Options
+    assert polymera.hamming(seq1, seq2, comparison="options") == 3
+
+    # Uncertainty
+    polymera.hamming(seq1, seq2, comparison="uncertainty")
+
+    # Wrong parameter
+    with pytest.raises(ValueError):
+        polymera.hamming(seq1, seq2, comparison="wrong_parameter")
+
+    # Unequal length
+    seq2.add_sequence_from_string("GGG")
+    with pytest.raises(ValueError):
+        polymera.hamming(seq1, seq2, comparison="options")