refactored _is_inversion

hyraxbio · Mar 9, 2017 · 24bb00d · 24bb00d
1 parent 03e63c4
commit 24bb00d
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 5 deletions.
diff --git a/seq2simulate/diversity.py b/seq2simulate/diversity.py
@@ -374,11 +374,13 @@ def _simulate_inversions(sequences, freq=0.3, max_length=200, min_length=15):
             while new_stops_introduced:
                 ins_start = random.randrange(0, len(seq)-min_length)
                 ins_length = random.randint(min_length, min(max_length, len(seq)-ins_start))
-                new_sequence = seq[0:ins_start] + seq[ins_start:ins_start + ins_length][::-1] + seq[ins_start + ins_length:]
+                inverted_sequence = seq[ins_start:ins_start + ins_length][::-1]
+                new_sequence = seq[0:ins_start] + inverted_sequence + seq[ins_start + ins_length:]
                 if _get_n_stop_codons(new_sequence) <= _get_n_stop_codons(seq):
                     new_stops_introduced = False
             sequences[i] = new_sequence
-            seq_diffs.append([ins_start, ins_start + ins_length])
+            seq_diffs.append([ins_start, ins_start + ins_length, inverted_sequence])
+            print(seq_diffs)
         else:
             seq_diffs.append([])
     return sequences, seq_diffs

diff --git a/seq2simulate/run_make_proviral_data.py b/seq2simulate/run_make_proviral_data.py
@@ -164,7 +164,7 @@ def _make_mutation_data_files(sequences, working_dir, hypermutation_rate=3):
                       'insertion': [diversity._simulate_insertions, {'freq': 0, 'no_frameshifts': True}], 
                       'frameshift': [diversity._simulate_frameshifts, {'freq': 0}],
                       'stopcodon': [diversity._simulate_stop_codons, {'freq': 0}],
-                      'inversion': [diversity._simulate_inversions, {'freq': 0}],
+                      'inversion': [diversity._simulate_inversions, {'freq': 1}],
                      }
 
     for i_mutation, mutation_type in enumerate(mutation_types):
@@ -332,8 +332,8 @@ def _is_stopcodon(sam_line, seq_diffs):
 
 def _is_inversion(sam_line, seq_diffs):
     if seq_diffs:
-        if seq_diffs[0] <= sam_line['read_start'] and seq_diffs[1] > sam_line['read_start'] \
-            or sam_line['read_start'] <= seq_diffs[0] <= sam_line['read_end']:
+        if _covers_index(seq_diffs[0], seq_diffs[1], sam_line['read_start']) \
+            or _covers_index(sam_line['read_start'], sam_line['read_end'], seq_diffs[0]):
             return True
     return False