fixed sliding window test

Author: LKremer

ursgal/resources/platform_independent/arc_independent/combine_pep_1_0_0/combine_pep_1_0_0.py

@@ -62,9 +62,10 @@ def sliding_window(elements, window_size, flexible=True):
     Gives you sliding window functionality without using container
     types (list, deque etc.) to speed it up. Only works for lists of
     numbers. Yields the sum of all numbers in the sliding window
-    (= the number of decoys in the sliding window in our case), and
-    the current length of the sliding window (= total number of PSMs
-    in the sliding window). Used for PEP calculation:
+    (= the number of decoys in the sliding window in our case), the
+    central number of the sliding window (required for the test only),
+    and the current length of the sliding window (= total number of
+    PSMs in the sliding window). Used for PEP calculation:
     PEP_of_PSM = (n_decoys_in_window * 2) / n_total_PSMs_in_window
     '''
     if flexible:
@@ -112,7 +113,7 @@ def sliding_window(elements, window_size, flexible=True):
                     break  # cause StopIteration silently ends for-loops, will be fixed in py3.6 :)
 
         previous_start_i, previous_stop_i = start_i, stop_i
-        yield n_decoys, current_win_size
+        yield n_decoys, center_value, current_win_size
 
 
 class CombinedPEP(object):
@@ -273,7 +274,7 @@ def generate_psm_to_scores_dict(self, input_engines):
                 kv_tuple[1]['Is decoy'] for kv_tuple in psms_sorted_by_bayes_pep
             ]
 
-            for i, (n_decoys, current_win_size) in enumerate(
+            for i, (n_decoys, __, current_win_size) in enumerate(
                 sliding_window(sorted_decoy_bools, self.window_size)):
 
                 n_false_positives = 2 * n_decoys