Merge pull request #70 from Nozdi/issue-63-gc-content

[#62] Added functions for gc content

src/shmir/data/ncbi_api.py

@@ -6,8 +6,8 @@
 import re
 
 from Bio import Entrez
-from designer import errors
-import settings
+from shmir.designer import errors
+from shmir import settings
 
 
 def get_data(

src/shmir/designer/validators.py

@@ -132,3 +132,15 @@ def check_input(seq_to_be_check):
             return check_complementary(ch_seq1[0], ch_seq2[0])
     else:
         raise errors.InputException('{}'.format(errors.error))
+
+
+def calculate_gc_content(sequence):
+    sequence = sequence.upper()
+    g_count = sequence.count('G')
+    c_count = sequence.count('C')
+
+    return int((float(g_count + c_count) / len(sequence)) * 100)
+
+
+def validate_gc_content(sequence, min_percent, max_percent):
+    return min_percent <= calculate_gc_content(sequence) <= max_percent

src/shmir/tests/test_ncbi_api.py

@@ -6,8 +6,8 @@
     patch
 )
 
-from data import ncbi_api
-from designer import errors
+from shmir.data import ncbi_api
+from shmir.designer import errors
 
 
 class TestGetDataFromNcbi(unittest.TestCase):

src/shmir/tests/test_validators.py

@@ -0,0 +1,27 @@
+import unittest
+
+from shmir.designer.validators import (
+    calculate_gc_content,
+    validate_gc_content,
+)
+
+
+class TestValidators(unittest.TestCase):
+
+    def test_calculate_gc_content(self):
+        seq_percent = calculate_gc_content('ACGT')
+        self.assertEqual(seq_percent, 50)
+
+        seq_percent = calculate_gc_content('CCCCGGGG')
+        self.assertEqual(seq_percent, 100)
+
+        seq_percent = calculate_gc_content('ACTTTTTTTA')
+        self.assertEqual(seq_percent, 10)
+
+    def test_validate_gc_content(self):
+        is_in_range = validate_gc_content('ACGT', 40, 60)
+        self.assertTrue(is_in_range)
+
+    def test_validate_gc_content_not_in_range(self):
+        not_in_range = validate_gc_content('ACGT', 52, 70)
+        self.assertFalse(not_in_range)