Merge pull request #73 from Eeeeelias/master

Fix multiple bugs in DDI extension part
daisybio · Sep 30, 2024 · f992b53 · f992b53
2 parents ea38381 + 3099c4b
commit f992b53
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 20 deletions.
diff --git a/preprocess/main_ddi_extend.py b/preprocess/main_ddi_extend.py
@@ -18,6 +18,9 @@
         print("This script only works with linux systems")
         exit(1)
     missing_files = []
+    if not os.path.isfile("database_sources.yml"):
+        print("No database_sources.yml file found, please read the README.md in sourcedata for more information.")
+        sys.exit(1)
     for file in needed_files:
         if not os.path.isfile(f"sourcedata/{file}"):
             missing_files.append(file)
@@ -26,7 +29,7 @@
         sys.exit(1)
 
     # parse yaml file with information about sources
-    tasks, organism, functions, additional_flags = parse_yaml.parse("sourcedata/database_sources.yml")
+    tasks, organism, functions, additional_flags = parse_yaml.parse("database_sources.yml")
 
     # purely debug
     if 'none' in functions:

diff --git a/preprocess/predict_interactions/parse_yaml.py b/preprocess/predict_interactions/parse_yaml.py
@@ -84,33 +84,33 @@ def parse_general(parse_sources, counter=1, col_name=None):
     if col_name is None:
         return tasks, counter
 
-    for source in parse_sources[col_name]:
-        try:
-            name = parse_sources[source]['name']
-        except KeyError:
-            name = source
+    source = col_name
+    try:
+        name = parse_sources[col_name]['name']
+    except KeyError:
+        name = source
 
-        try:
-            path = parse_sources[source]['path']
-        except KeyError:
-            print(f"YAML not properly formatted, specify a path for mippie")
-            exit(1)
+    try:
+        path = parse_sources[col_name]['path']
+    except KeyError:
+        print(f"YAML not properly formatted, specify a path for mippie")
+        exit(1)
 
-        try:
-            mapping = parse_sources[source]['mapping']
-        except KeyError:
-            print(f"YAML not properly formatted, specify mapping for {name}")
-            exit(1)
-        tasks.append((f"clean_{source}", path, f"sourcedata/source{counter}_{name}", mapping))
-        counter += 1
+    try:
+        mapping = parse_sources[col_name]['mapping']
+    except KeyError:
+        print(f"YAML not properly formatted, specify mapping for {name}")
+        exit(1)
+    tasks.append((f"clean_{source}", path, f"sourcedata/source{counter}_{name}", mapping))
+    counter += 1
     return tasks, counter
 
 
 # feel free to extend this list by adding a new function and updating the dictionary
 supported_sources = {'mitab': parse_mitab,
                      'string': parse_string,
-                     'mippie': parse_general,
-                     'homology': parse_general,
+                     'mippie': parse_mippie,
+                     'homology': parse_homology,
                      'mint': parse_mint}
 
 
@@ -131,6 +131,7 @@ def parse(file):
         if supported_source not in sources_to_parse:
             continue
         source_tasks, counter = supported_sources[supported_source](sources_to_parse, counter)
+        tasks.extend(source_tasks)
 
     if 'params' in data:
         for param in data['params']: