add strandedness to the sequence fetch

ensembl/core/base.py

@@ -1,4 +1,5 @@
 from django.db import models
+from ensembl.utils import reverse_complement
 
 
 class StableIdManager(models.Manager):
@@ -38,6 +39,7 @@ class Meta:
     seq_region = models.ForeignKey('SeqRegion')
     seq_region_start = models.IntegerField()
     seq_region_end = models.IntegerField()
+    seq_region_strand = models.IntegerField()
 
     def sequence_level_assemblies(self):
         """return the set of assemblies at the sequence level for this object"""
@@ -62,10 +64,20 @@ def projected_coords(self, assemblies):
     @property 
     def sequence(self):
         """return the sequence for this object"""
+
+        # fetch the component assemblies and their DNA and project coords into those
         components = self.sequence_level_assemblies().select_related('cmp_seq_region__dna').all()
         coords = self.projected_coords(components)
-        return ''.join([
+
+        # build the sequence from the component parts
+        sequence = ''.join([
             component.cmp_seq_region.dna.sequence[start:end]
             for (component,(start, end)) in zip(components, coords)
         ])
+
+        # reverse complement if necessary
+        if self.seq_region_strand == -1: 
+            sequence = reverse_complement(sequence)
+
+        return sequence
 

ensembl/core/models.py

@@ -178,7 +178,7 @@ class Exon(HasSeqRegion, HasStableId):
     # seq_region_id = models.IntegerField()
     # seq_region_start = models.IntegerField()
     # seq_region_end = models.IntegerField()
-    seq_region_strand = models.IntegerField()
+    # seq_region_strand = models.IntegerField()
     phase = models.IntegerField()
     end_phase = models.IntegerField()
     is_current = models.IntegerField()
@@ -241,7 +241,7 @@ class Gene(HasSeqRegion, HasStableId):
     # seq_region_id = models.IntegerField()
     # seq_region_start = models.IntegerField()
     # seq_region_end = models.IntegerField()
-    seq_region_strand = models.IntegerField()
+    # seq_region_strand = models.IntegerField()
     display_xref_id = models.IntegerField(null=True, blank=True)
     source = models.CharField(max_length=60)
     status = models.CharField(max_length=57, blank=True)
@@ -653,7 +653,7 @@ class Transcript(HasSeqRegion, HasStableId):
     # seq_region_id = models.IntegerField()
     # seq_region_start = models.IntegerField()
     # seq_region_end = models.IntegerField()
-    seq_region_strand = models.IntegerField()
+    # seq_region_strand = models.IntegerField()
     display_xref_id = models.IntegerField(null=True, blank=True)
     biotype = models.CharField(max_length=120)
     status = models.CharField(max_length=57, blank=True)

ensembl/core/routers.py

@@ -1,10 +1,10 @@
 class EnsemblRouter(object):
+    """ Base class for routing, allows reads and prevents everything else"""
 
     app = None
     database = None
 
     def db_for_read(self, model, **hints):
-        print model, model._meta.app_label
         if model._meta.app_label == self.app:
             return self.database
         return None

ensembl/utils.py

@@ -1,5 +1,5 @@
 import unittest
-from django.db import models
+import string
 
 
 class BasicTest(object):
@@ -28,3 +28,7 @@ def test_sequence_lookup_returns_object_of_correct_length(self):
     # def test_sequence_lookup_returns_object_in_correct_orientation(self):
     #     assert(False)
 
+
+_rc_trans = string.maketrans('ACGTN', 'TGCAN')
+def reverse_complement(seq):
+    return string.translate(str(seq), _rc_trans)[::-1]

ensembl/variation/models.py

@@ -243,7 +243,7 @@ class VariationFeature(HasSeqRegion):
     # seq_region_id = models.IntegerField()
     # seq_region_start = models.IntegerField()
     # seq_region_end = models.IntegerField()
-    seq_region_strand = models.IntegerField()
+    # seq_region_strand = models.IntegerField()
     variation = models.ForeignKey('Variation')
     allele_string = models.TextField(blank=True)
     variation_name = models.CharField(max_length=765, blank=True)