Auto-download complex and family data

geneontology · Apr 28, 2020 · 67e915f · 67e915f
1 parent 0b96894
commit 67e915f
Show file tree

Hide file tree

Showing 3 changed files with 49 additions and 8 deletions.
diff --git a/download.py b/download.py
@@ -0,0 +1,28 @@
+import requests
+import os
+from requests.models import Response
+
+
+class SignorDownloader:
+    @staticmethod
+    def write_response_content(response: Response, file_basename, destination_dir=None):
+        if destination_dir is None:
+            destination_dir = "resources"
+        file_target = os.path.join(destination_dir, file_basename)
+        with open(file_target, "wb") as ft:
+            ft.write(response.content)
+        return file_target
+
+    @staticmethod
+    def download_complexes(destination_dir=None):
+        response = requests.post(url="https://signor.uniroma2.it/download_complexes.php",
+                                 data={"submit": "Download complex data"})
+        file_basename = "SIGNOR_complexes.csv"
+        return SignorDownloader.write_response_content(response, file_basename, destination_dir=destination_dir)
+
+    @staticmethod
+    def download_families(destination_dir=None):
+        response = requests.post(url="https://signor.uniroma2.it/download_complexes.php",
+                                 data={"submit": "Download protein family data"})
+        file_basename = "SIGNOR_PF.csv"
+        return SignorDownloader.write_response_content(response, file_basename, destination_dir=destination_dir)
diff --git a/entity_factories.py b/entity_factories.py
@@ -1,6 +1,8 @@
 import csv
+import os
 from naming_conventions import NamingConvention
 from entity_models import SignorEntity, SignorProtein, SignorComplex, SignorProteinFamily
+from download import SignorDownloader
 
 
 class SignorGroupingFactory:
@@ -28,26 +30,38 @@ def __init__(self, filename):
 
 
 class SignorComplexFactory(SignorGroupingFactory):
-    def __init__(self, filename):
+    FILENAME = "resources/SIGNOR_complexes.csv"
+
+    def __init__(self, filename=None):
+        if filename is None:
+            filename = self.FILENAME
+        if not os.path.exists(filename):
+            filename = SignorDownloader.download_complexes()
+
         self.NAME_FIELD = "COMPLEX NAME"
         self.GROUPING_CLASS = SignorComplex
         SignorGroupingFactory.__init__(self, filename)
         self.complexes = self.grouping
 
 
 class SignorProteinFamilyFactory(SignorGroupingFactory):
-    def __init__(self, filename):
+    FILENAME = "resources/SIGNOR_PF.csv"
+
+    def __init__(self, filename=None):
+        if filename is None:
+            filename = self.FILENAME
+        if not os.path.exists(filename):
+            filename = SignorDownloader.download_families()
+
         self.NAME_FIELD = "PROT. FAMILY NAME"
         self.GROUPING_CLASS = SignorProteinFamily
         SignorGroupingFactory.__init__(self, filename)
         self.families = self.grouping
 
 
 class SignorEntityFactory:
-    complex_csv_filename = "SIGNOR_complexes.csv"
-    complex_factory = SignorComplexFactory(complex_csv_filename)
-    family_csv_filename = "SIGNOR_PF.csv"
-    family_factory = SignorProteinFamilyFactory(family_csv_filename)
+    complex_factory = SignorComplexFactory()
+    family_factory = SignorProteinFamilyFactory()
 
     @classmethod
     def determine_entity(cls, entity_id: str, entity_name: str) -> SignorEntity:

diff --git a/pathway_connections.py b/pathway_connections.py
@@ -6,8 +6,7 @@
 from rdflib.term import URIRef
 from prefixcommons.curie_util import expand_uri
 from entity_factories import SignorEntityFactory
-from naming_conventions import NamingConvention
-from entity_models import SignorEntity, SignorProtein, SignorComplex
+from entity_models import SignorEntity
 
 ro = OboRO()