Merge pull request #88 from MaozGelbart/pssmmotif_from_sequences

MAINT: adapt MotifPssmPattern to modern biopython

dnachisel/SequencePattern/MotifPssmPattern.py

@@ -16,7 +16,7 @@ class MotifPssmPattern(SequencePattern):
     ----------
 
     pssm
-      A Bio.Align.AlignInfo.PSSM object.
+      A `Bio.motifs.Motif` object.
 
     threshold
      locations of the sequence with a PSSM score above this value will be
@@ -32,6 +32,10 @@ class MotifPssmPattern(SequencePattern):
     def __init__(
         self, pssm, threshold=None, relative_threshold=None,
     ):
+        if not isinstance(pssm, Bio.motifs.Motif):
+            raise ValueError(
+                f"Expected PSSM type of `Bio.motifs.Motif`, but {type(pssm)} was passed"
+            )
         self.name = pssm.name
         self.pssm = pssm.pssm
         if relative_threshold is not None:
@@ -108,8 +112,8 @@ def from_sequences(
         sequences = [Seq(s) for s in sequences]
         motif = motifs.create(sequences)
         cls.apply_pseudocounts(motif, pseudocounts)
-        pssm = PSSM(motif.pssm)
-        pssm.name = name
+        motif.name = name
+        pssm = motif
         return MotifPssmPattern(
             pssm=pssm, threshold=threshold, relative_threshold=relative_threshold,
         )

tests/test_patterns.py

@@ -37,3 +37,12 @@ def test_pssm_pattern_from_file(
     )
     assert len(multiple_patterns) == 2
     assert all([isinstance(p, MotifPssmPattern) for p in multiple_patterns])
+
+
+def test_pssm_from_sequences():
+    seqs = ['ACGT', 'ACCT', 'AAGT']
+    motif_name = "test motif"
+    pat = MotifPssmPattern.from_sequences(seqs, name=motif_name, relative_threshold=0.9)
+    assert isinstance(pat, MotifPssmPattern)
+    assert len(pat.find_matches_in_string("ACATACGTACAC")) == 1
+    assert len(pat.find_matches_in_string("ACATAATTACAC")) == 0