Merge branch 'medvir:master' into master

src/minvar/annotate.py

@@ -50,10 +50,10 @@
         os.sys.path.insert(1, dn_dir)
         mod = __import__('minvar')
         sys.modules["minvar"] = mod
-        from common import h77_map, consensus_B_map, d2a, coverage_threshold
+        from common import h77_map, consensus_B_map, d2a, coverage_threshold, AMBIGUITY_THRESHOLD
         from Alignment import needle_align, alignfile2dict
 else:
-    from .common import h77_map, consensus_B_map, d2a, coverage_threshold
+    from .common import h77_map, consensus_B_map, d2a, coverage_threshold, AMBIGUITY_THRESHOLD
     from .Alignment import needle_align, alignfile2dict
 
 
@@ -416,7 +416,7 @@ def df_2_ambiguous_sequence(df_in, cov_df=None):
     """
     assert 'freq' in df_in.columns
     # select calls with freq > 15%
-    df_in = df_in[df_in['freq'] >= 0.15]
+    df_in = df_in[df_in['freq'] >= AMBIGUITY_THRESHOLD]
     # aggregate calls for the same position
     all_nt = df_in.groupby(['pos']).agg({'mut': lambda x: ''.join(sorted(x))})
     # create a columng of ambiguous bases * so insertion is not considered because only the first nc is returned

src/minvar/common.py

@@ -9,6 +9,8 @@
 MIN_FRACTION = 0.015
 RAW_DEPTH_THRESHOLD = 50
 coverage_threshold = 100
+AMBIGUITY_THRESHOLD = 0.15
+APD_THRESHOLD = 0.01
 
 # This must be updated when new masterComments files are downloaded into src/minvar/db/HIV
 mastercomments_version = 'HIVdb version 8.4 (last updated 2017-06-16)'