Biopython 1.76 Compatibility and Alphabet Removal

pssm_model.py and binding_model.py updated to run biopython 1.76 with the
removal of the alphabet function. Alphabet added to parent class
binding_model.py for the use of pssm_model.py

cgb/binding_model.py

@@ -33,15 +33,21 @@ class TFBindingModel():
     __metaclass__ = ABCMeta
 
     def __init__(self, collections,
-                 background={'A': 0.25, 'C': 0.25, 'G': 0.25, 'T': 0.25}):
+                 background={'A': 0.25, 'C': 0.25, 'G': 0.25, 'T': 0.25}, alphabet="ACGT"):
         """Constructor for the TFBindingModel class.
 
         Combines multiple PSWMs using phylogenetic weighting to derive a
         species-specific PWM.
         """
         self._background = background
         self._collection_set = collections
-
+        self._alphabet = alphabet
+
+    @cached_property
+    def alphabet(self):
+        """Returns the DNA alphabet to be used by pssm_model.py"""
+        return self._alphabet
+
     @cached_property
     def background(self):
         """Returns the background distribution of the model."""

cgb/pssm_model.py

@@ -29,10 +29,10 @@ class PSSMModel(TFBindingModel):
     target species.
     """
     def __init__(self, collections, weights,
-                 background={'A': 0.25, 'C': 0.25, 'G': 0.25, 'T': 0.25}):
+                 background={'A': 0.25, 'C': 0.25, 'G': 0.25, 'T': 0.25}, alphabet="ACGT"):
         """Constructor for the PSSMModel class."""
-        super(PSSMModel, self).__init__(collections, background)
-        self._pwm = self._combine_pwms([c.pwm for c in collections], weights)
+        super(PSSMModel, self).__init__(collections, background, alphabet)
+        self._pwm = self._combine_pwms([c.pwm for c in collections], weights, alphabet)
 
     @cached_property
     def pwm(self):
@@ -53,12 +53,7 @@ def pssm(self):
     def rev_comp_pssm(self):
         """Returns the reverse complement of the PSSM."""
         return self.pssm.reverse_complement()
-
-    @property
-    def alphabet(self):
-        """Returns the alphabet of the motif."""
-        return self.pwm.alphabet
-
+
     @cached_property
     def IC(self):
         """Returns the information content of the PSSM."""
@@ -140,7 +135,7 @@ def score_seq(self, seq, both=True):
     @property
     def weblogo_from_pwm(self):
         s = len(self.sites)
-        alpha = self.alphabet.letters
+        alpha = self._alphabet
         cols = []
         for i in range(self.length):
             cols.append("")
@@ -162,16 +157,13 @@ def weblogo_from_pwm(self):
         return filename
 
     @staticmethod
-    def _combine_pwms(pwms, weights):
+    def _combine_pwms(pwms, weights, alphabet):
         """Combines the given PWMs according to the given weights."""
         len = pwms[0].length
-        alphabet = pwms[0].alphabet
         # Check if all PWMs are of the same length.
         assert all(len == pwm.length for pwm in pwms)
-        # Check if all PWMs have the same alphabet -- 'ACGT'
-        assert all(alphabet == pwm.alphabet for pwm in pwms)
         # Combine all PWMs according to given weights
         pwm_vals = {let: [sum(pwm[let][i]*w for pwm, w in zip(pwms, weights))
                           for i in xrange(len)]
-                    for let in alphabet.letters}
+                    for let in alphabet}  
         return PositionWeightMatrix(alphabet, pwm_vals)

conda_cgb_environment.yml

@@ -10,7 +10,7 @@ dependencies:
   - scipy=1.2.1
   - matplotlib=2.2.5
   - pyparsing=2.4.7
-  - biopython=1.73
+  - biopython=1.76
   - pyzmq=19.0.0
   - ipython=5.8.0
   - spyder=4.1.1
@@ -21,7 +21,7 @@ dependencies:
   - pip
   - pip:
     # works for regular pip packages
-    - ete3=3.1.1
-    - tqdm=4.47.0
+    - ete3==3.1.1
+    - tqdm==4.47.0
     - weblogo==3.4