Change rules around missing coverage to match ruby version for #412.

cfe-lab · Feb 2, 2018 · 1ea6d4b · 1ea6d4b
1 parent 90c7bac
commit 1ea6d4b
Show file tree

Hide file tree

Showing 2 changed files with 68 additions and 16 deletions.
diff --git a/micall/hivdb/hivdb.py b/micall/hivdb/hivdb.py
@@ -336,25 +336,52 @@ def interpret(asi, amino_seq, region):
     is_missing_midi = region.endswith('-NS5b') and len(amino_seq) < len(ref_seq)
 
     if is_missing_midi:
-        # At least some data found, assume wild type in gaps.
-        new_amino_seq = []
-        for wild_type, old_aminos in zip_longest(ref_seq, amino_seq):
-            if old_aminos is not None:
-                new_amino_seq.append(old_aminos)
-            else:
-                new_amino_seq.append({wild_type: 1.0})
-        amino_seq = new_amino_seq
-
+        amino_seq += [{}] * (len(ref_seq) - len(amino_seq))
     result = asi.interpret(amino_seq, region)
+
     if not is_missing_midi:
-        return result
+        for drug_result in result.drugs:
+            if drug_result.level == ResistanceLevels.FAIL.level:
+                break
+        else:
+            # No missing coverage, we're done.
+            return result
+
+    # At least some data found, assume wild type in gaps.
+    new_amino_seq = []
+    for wild_type, old_aminos in zip_longest(ref_seq, amino_seq):
+        if old_aminos:
+            new_amino_seq.append(old_aminos)
+        else:
+            new_amino_seq.append({wild_type: 1.0})
+    amino_seq = new_amino_seq
+
+    new_result = asi.interpret(amino_seq, region)
+    new_drug_results = {drug_result.code: drug_result
+                        for drug_result in new_result.drugs}
+    if is_missing_midi:
+        allowed_levels = (ResistanceLevels.FAIL.level,
+                          ResistanceLevels.RESISTANCE_LIKELY.level,
+                          ResistanceLevels.NOT_INDICATED.level)
+    else:
+        allowed_levels = (ResistanceLevels.FAIL.level,
+                          ResistanceLevels.UNKNOWN_MUTATIONS.level,
+                          ResistanceLevels.RESISTANCE_POSSIBLE.level,
+                          ResistanceLevels.RESISTANCE_LIKELY.level,
+                          ResistanceLevels.NOT_INDICATED.level)
+
     for drug_result in result.drugs:
-        if drug_result.level not in (ResistanceLevels.FAIL.level,
-                                     ResistanceLevels.RESISTANCE_LIKELY.level,
-                                     ResistanceLevels.NOT_INDICATED.level):
-            drug_result.level = ResistanceLevels.FAIL.level
-            drug_result.level_name = ResistanceLevels.FAIL.name
-            drug_result.score = 0.0
+        if drug_result.level == ResistanceLevels.FAIL.level or is_missing_midi:
+            new_drug_result = new_drug_results[drug_result.code]
+            if new_drug_result.level in allowed_levels:
+                drug_result.level = new_drug_result.level
+                drug_result.level_name = new_drug_result.level_name
+                drug_result.score = new_drug_result.score
+            else:
+                drug_result.level = ResistanceLevels.FAIL.level
+                drug_result.level_name = ResistanceLevels.FAIL.name
+                drug_result.score = 0.0
+    result.mutations = new_result.mutations
     return result
 
 

diff --git a/micall/tests/test_hivdb.py b/micall/tests/test_hivdb.py
@@ -1131,6 +1131,31 @@ def test_hcv_missing_midi(self):
         self.assertEqual(expected_resistance, resistance_csv.getvalue())
         self.assertEqual(expected_mutations, mutations_csv.getvalue())
 
+    def test_hcv_ns3_coverage_gap(self):
+        self.maxDiff = None
+        ns3_aminos = [{c: 1.0}
+                       for c in self.algorithms[None].stds['HCV1A-H77-NS3']]
+        ns3_aminos[53] = {}  # Coverage gap at position 54
+        aminos = [AminoList('HCV1A-H77-NS3', ns3_aminos, '1A')]
+        resistance_csv = StringIO()
+        mutations_csv = StringIO()
+        expected_resistance = """\
+region,drug_class,drug,drug_name,level,level_name,score,genotype
+NS3,NS3,BPV,BPV,0,Sequence does not meet quality-control standards,0.0,1A
+NS3,NS3,GZR,GZR,1,Likely Susceptible,0.0,1A
+NS3,NS3,PTV,PTV,1,Likely Susceptible,0.0,1A
+NS3,NS3,SPV,SPV,0,Sequence does not meet quality-control standards,0.0,1A
+NS3,NS3,TPV,TPV,0,Sequence does not meet quality-control standards,0.0,1A
+"""
+        expected_mutations = """\
+drug_class,mutation,prevalence,genotype
+"""
+
+        write_resistance(aminos, resistance_csv, mutations_csv, self.algorithms)
+
+        self.assertEqual(expected_resistance, resistance_csv.getvalue())
+        self.assertEqual(expected_mutations, mutations_csv.getvalue())
+
     def test_hcv_low_coverage(self):
         aminos = []
         resistance_csv = StringIO()