Closes #5

.gitlab-ci.yml

@@ -6,11 +6,11 @@ cache:
                 - balibase/
 test:
         script:
-                - python3 -m unittest alignementseq_tests test_alignementseq_multiple
+                - python3 -m unittest alignementseq_tests test_alignementseq_multiple -v
 
 test_performance:
         script:
-                - python3 -m unittest alignementseq_tests.PerformanceTestCase
+                - python3 -m unittest alignementseq_tests.PerformanceTestCase -v
         artifacts:
                 paths:
                         - timings.png

alignementseq.py

@@ -164,20 +164,23 @@ def vec_align(seq1, seq2, mat=dict_to_weight(Bio.SubsMat.MatrixInfo.blosum62), g
     S[0][0] = 0
     S[1][0] = gap
     S[1][1] = gap
+
+    x *= 26
     for s in range(2,n+m+1):
         # First/last cell must be treated differently when first line/col of M
         if(s <= m):
             S[s][0] = S[s-1][0] + gap
         if(s <= n):
             S[s][s-max(0,s-m)] = S[s-1][s-1-max(0,s-1-m)] + gap
+
         # The offsets in the formula become begin/end of slices.
+        gapped = S[s-1] + gap
         S[s][int(s<=m):len(S[s])-int(s<=n)] = np.maximum(np.maximum(
-            S[s-1][1:] + gap,
-            S[s-1][:-1] + gap),
+            gapped[1:],
+            gapped[:-1]),
             S[s-2][int(s>m+1):len(S[s-2])-int(s>n+1)]
             +mat[
-                x[max(0,s-m-1):min(n,s-1)] * 26 +
-                y[max(0,s-n-1):min(m,s-1)][::-1]
+                x[max(0,s-m-1):min(n,s-1)] + y[max(0,s-n-1):min(m,s-1)][::-1]
                 ]
             )
 

alignementseq_tests.py

@@ -19,11 +19,10 @@ import Bio.Entrez
 from Bio.SeqRecord import SeqRecord
 from Bio.SubsMat.MatrixInfo import blosum62
 
-class AlignmentSeqTestCase(unittest.TestCase):
+class BalibaseTestCase(unittest.TestCase):
+    """Base class to inherit by all test cases that need access to the balibase.
+    """
     def setUp(self):
-        
-        self.unit_timeout = int(os.environ.get("UNIT_TIMEOUT") or 60)
-        
         balibase_zippath = "balibase.zip"
         self.balibase_path = "balibase"
         testfile_path = os.path.join(
@@ -53,6 +52,8 @@ class AlignmentSeqTestCase(unittest.TestCase):
             seq1 = next(records)
             seq2 = next(records)
             yield seq1, seq2, filename
+
+class AlignmentSeqTestCase(BalibaseTestCase):
     
     def get_dataset_records(self):
         """Generator function to iterate over the record generator  of each
@@ -212,7 +213,7 @@ class AlignmentSeqTestCase(unittest.TestCase):
             Bio.SeqIO.write((r1, r2), fd, "fasta")
     
 
-class PerformanceTestCase(AlignmentSeqTestCase):
+class PerformanceTestCase(BalibaseTestCase):
     """Performance tests, slow.
 
     Excluded from default test suite, run it with
@@ -224,6 +225,8 @@ class PerformanceTestCase(AlignmentSeqTestCase):
         """
         super().setUp()
 
+        self.unit_timeout = int(os.environ.get("UNIT_TIMEOUT") or 60)
+
         with warnings.catch_warnings():
             warnings.simplefilter("ignore")
             self.titin_human, self.titin_mouse = Bio.SeqIO.parse(