review changes

big_scape/comparison/extend.py

@@ -596,7 +596,7 @@ def extend_greedy(pair: RecordPair) -> None:
 
 def extend_simple_match(pair: RecordPair, match, gap):
     """Performs extension by first creating a simple match matrix, then
-    performing a match/gap extentsion similar to legaxy expansion
+    performing a match/gap extentsion similar to legacy expansion
 
     This method expects LCS to have been performed on the pair, and will
     do all four directions at once
@@ -612,11 +612,11 @@ def extend_simple_match(pair: RecordPair, match, gap):
     b_domains = []
 
     # so we'll do a loop through cds and through domains to keep track of everything
-    for cds_idx, cds in enumerate(pair.record_a.get_cds()):
+    for cds_idx, cds in enumerate(pair.record_a.get_cds_with_domains()):
         for domain in cds.hsps:
             a_domains.append((domain, cds_idx))
 
-    for cds_idx, cds in enumerate(pair.record_b.get_cds()):
+    for cds_idx, cds in enumerate(pair.record_b.get_cds_with_domains()):
         for domain in cds.hsps:
             b_domains.append((domain, cds_idx))
 
@@ -632,7 +632,7 @@ def extend_simple_match(pair: RecordPair, match, gap):
     # a forward
     score = 0
     max_score = 0
-    for i in range(pair.comparable_region.lcs_a_stop + 1, len(a_domains)):
+    for i in range(pair.comparable_region.lcs_domain_a_stop, len(a_domains)):
         domain = a_domains[i][0]
 
         if domain not in common_domains:
@@ -643,13 +643,13 @@ def extend_simple_match(pair: RecordPair, match, gap):
         if score > max_score:
             cds_idx = a_domains[i][1]
             max_score = score
-            pair.comparable_region.a_stop = cds_idx
-            pair.comparable_region.domain_a_stop = i
+            pair.comparable_region.a_stop = cds_idx + 1
+            pair.comparable_region.domain_a_stop = i + 1
 
     # a reverse
     score = 0
     max_score = 0
-    for i in range(pair.comparable_region.lcs_a_start - 1, -1, -1):
+    for i in range(pair.comparable_region.lcs_domain_a_start - 1, -1, -1):
         domain = a_domains[i][0]
 
         if domain not in common_domains:
@@ -666,7 +666,7 @@ def extend_simple_match(pair: RecordPair, match, gap):
     # b forward
     score = 0
     max_score = 0
-    for i in range(pair.comparable_region.lcs_b_stop + 1, len(b_domains)):
+    for i in range(pair.comparable_region.lcs_domain_b_stop, len(b_domains)):
         domain = b_domains[i][0]
 
         if domain not in common_domains:
@@ -677,13 +677,13 @@ def extend_simple_match(pair: RecordPair, match, gap):
         if score > max_score:
             cds_idx = b_domains[i][1]
             max_score = score
-            pair.comparable_region.b_stop = cds_idx
-            pair.comparable_region.domain_b_stop = i
+            pair.comparable_region.b_stop = cds_idx + 1
+            pair.comparable_region.domain_b_stop = i + 1
 
     # b reverse
     score = 0
     max_score = 0
-    for i in range(pair.comparable_region.lcs_b_start - 1, -1, -1):
+    for i in range(pair.comparable_region.lcs_domain_b_start - 1, -1, -1):
         domain = b_domains[i][0]
 
         if domain not in common_domains:

big_scape/enums/comparison.py

@@ -14,6 +14,7 @@ class ALIGNMENT_MODE(Enum):
 class EXTEND_STRATEGY(Enum):
     LEGACY = "legacy"
     GREEDY = "greedy"
+    SIMPLE_MATCH = "simple_match"
 
 
 class COMPARISON_MODE(Enum):

test/comparison/test_extend.py

@@ -907,13 +907,13 @@ def test_match_extend(self):
 
         # emulate LCS
         pair.comparable_region = bs_comp.ComparableRegion(
-            9, 10, 6, 7, 9, 10, 6, 7, False
+            9, 11, 6, 8, 9, 11, 6, 8, False
         )
 
         bs_comp.extend.extend_simple_match(pair, 5, -2)
 
         expected_comparable_region = bs_comp.ComparableRegion(
-            6, 10, 5, 7, 6, 10, 5, 7, False
+            6, 11, 5, 8, 6, 11, 5, 8, False
         )
 
         self.assertEqual(pair.comparable_region, expected_comparable_region)
@@ -933,3 +933,44 @@ def test_match_extend(self):
             pair.comparable_region.domain_b_stop,
             expected_comparable_region.domain_b_stop,
         )
+
+    def test_expand_simple_match_multi_domain(self):
+        """Tests glocal expand with multi domain cdss"""
+        # brackets indicate a cds with multiple domains
+        #
+        # A:        [XX]BX[A BC]DEX XXXX
+        # B: XXXXXXXXXX XX A[BC]EDX[XXXX]
+        #
+        a_cds, b_cds = generate_mock_cds_lists(
+            10, 17, [3, 3, 3, 4, 5], [12, 13, 13, 15, 14], False
+        )
+        a_cds[0].hsps.append(a_cds[0].hsps[0])
+        a_cds[1].hsps = [a_cds[3].hsps[1]]
+        b_cds[-1].hsps.extend([b_cds[-1].hsps[0]] * 3)
+        record_a = generate_mock_region(a_cds)
+        record_b = generate_mock_region(b_cds)
+        pair = big_scape.comparison.record_pair.RecordPair(record_a, record_b)
+        pair.comparable_region = bs_comp.ComparableRegion(
+            3, 4, 12, 14, 4, 7, 12, 15, False
+        )
+        bs_comp.extend.extend_simple_match(pair, 5, -2)
+        expected_comparable_region = bs_comp.ComparableRegion(
+            1, 6, 12, 16, 2, 9, 12, 17, False
+        )
+        self.assertEqual(pair.comparable_region, expected_comparable_region)
+        self.assertEqual(
+            pair.comparable_region.domain_a_start,
+            expected_comparable_region.domain_a_start,
+        )
+        self.assertEqual(
+            pair.comparable_region.domain_b_start,
+            expected_comparable_region.domain_b_start,
+        )
+        self.assertEqual(
+            pair.comparable_region.domain_a_stop,
+            expected_comparable_region.domain_a_stop,
+        )
+        self.assertEqual(
+            pair.comparable_region.domain_b_stop,
+            expected_comparable_region.domain_b_stop,
+        )