Fix for #51

loosolab · Jan 10, 2019 · 97c526d · 97c526d
1 parent a161ed0
commit 97c526d
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 6 deletions.
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/ActivityCategorizer.py b/bin/3.1_create_gtf/Modules/Ensembl/ActivityCategorizer.py
@@ -122,8 +122,9 @@ def activity_comparator(self, aliaslist):
         concatenated_array = bytearray([])
         length = len(self.activity[aliaslist[0]])
         input_arrays = [self.activity[index] for index in aliaslist]
-        try:
-            for x in range(length):
+        for x in range(length):
+            # This try-catch block is needed because of inconsistency in file-lengths in Ensembl-release-95
+            try:
                 if any(y[x] == 0 for y in input_arrays):
                     concatenated_array.append(0)
                 elif any(y[x] == 1 for y in input_arrays):
@@ -134,8 +135,9 @@ def activity_comparator(self, aliaslist):
                     concatenated_array.append(3)
                 elif any(y[x] == 4 for y in input_arrays):
                     concatenated_array.append(4)
-        except IndexError:
-            print("Indexerror occured")
+            except IndexError:
+                concatenated_array.append(4)
+
         return concatenated_array
 
 

diff --git a/bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py b/bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py
@@ -107,10 +107,13 @@ def generate_activity_list(self, activity, index):
 
         activity_list = []
         for key, value in activity.items():
+            #  if no index is found the key will be annotated as "NA"
+            #  this is needed due to an inconsistency in ensembl-release-95
+            #  some activity-files contain less entries than others
             try:
                 activity_list.append(key+">"+self.value_map[value[index]])
             except IndexError:
-                pass
+                activity_list.append(key+">NA")
         return activity_list
 
     def get_gtf(self, release, activity):

diff --git a/bin/3.1_create_gtf/config/celltypes_homo_sapiens.json b/bin/3.1_create_gtf/config/celltypes_homo_sapiens.json
@@ -129,7 +129,7 @@
     {
       "type": "Heart",
       "alias_ucsc": ["heart"],
-      "alias_ensembl": ["Right_Atrium", "Left_Ventricle", "cardiac_muscle_cell", "heart_right_ventricle"]
+      "alias_ensembl": ["Right_Atrium", "Left_Ventricle", "cardiac_muscle_cell", "heart_right_ventricle", "heart"]
     },
     {
       "type": "Osteoblast",