Merge branch 'master' of github.com:saezlab/pypath

src/pypath/core/network.py

@@ -431,7 +431,7 @@ def __init__(
             df_dtype = None,
             pickle_file = None,
             ncbi_tax_id = 9606,
-            allow_loops = True,
+            allow_loops = None,
             **kwargs
         ):
 
@@ -521,6 +521,7 @@ def load(
             cache_files = None,
             only_directions = False,
             pickle_file = None,
+            allow_loops = None,
         ):
         """
         Loads data from a network resource or a collection of resources.
@@ -553,6 +554,7 @@ def load(
             'keep_raw': keep_raw,
             'top_call': False,
             'only_directions': only_directions,
+            'allow_loops': allow_loops,
         }
 
         exclude = common.to_set(exclude)
@@ -622,6 +624,7 @@ def load_resource(
             redownload = None,
             keep_raw = False,
             only_directions = False,
+            allow_loops = None,
             **kwargs
         ):
         """
@@ -664,15 +667,24 @@ def load_resource(
             redownload = redownload,
             keep_raw = keep_raw,
         )
-        self._add_edge_list(only_directions = only_directions)
 
-        self.organisms_check()
-        self.remove_zero_degree()
+        allow_loops = self._allow_loops(
+            allow_loops = allow_loops,
+            resource = resource,
+        )
 
-        if not self.allow_loops:
+        self._log('Loops allowed for resource `%s`: %s' % (
+            resource.name,
+            allow_loops,
+        ))
 
-            self.remove_loops()
+        self._add_edge_list(
+            only_directions = only_directions,
+            allow_loops = allow_loops,
+        )
 
+        self.organisms_check()
+        self.remove_zero_degree()
 
         self._log(
             'Completed: loading network data from '
@@ -1664,6 +1676,7 @@ def _add_edge_list(
             edge_list = False,
             regulator = False,
             only_directions = False,
+            allow_loops = None,
         ):
         """
         Adds edges to the network from *edge_list* obtained from file or
@@ -1684,6 +1697,8 @@ def _add_edge_list(
 
         self._log('Adding preprocessed edge list to existing network.')
 
+        allow_loops = self._allow_loops(allow_loops = allow_loops)
+
         if not edge_list:
 
             if (
@@ -1713,11 +1728,15 @@ def _add_edge_list(
 
                 return False
 
-        edges = []
+        self._filtered_loops = 0
 
         for e in edge_list:
 
-            self._add_update_edge(e, only_directions = only_directions)
+            self._add_update_edge(
+                e,
+                allow_loops = allow_loops,
+                only_directions = only_directions,
+            )
 
         self._log(
             'New network resource added, current number '
@@ -1727,12 +1746,19 @@ def _add_edge_list(
             )
         )
 
+        if not allow_loops:
+
+            self._log('Loop edges discarded: %u' % self._filtered_loops)
+
+        delattr(self, '_filtered_loops')
+
         self.raw_data = None
 
 
     def _add_update_edge(
             self,
             edge,
+            allow_loops = None,
             only_directions = False,
         ):
         """
@@ -1817,6 +1843,8 @@ def _add_update_edge(
             edge['attrs_node_b'],
         )
 
+        allow_loops = allow_loops or self.allow_loops
+
         refs = {refs_mod.Reference(pmid) for pmid in refs}
 
         entity_a = entity_mod.Entity(
@@ -1839,6 +1867,11 @@ def _add_update_edge(
             b = entity_b,
         )
 
+        if not allow_loops and interaction.is_loop():
+
+            self._filtered_loops += 1
+            return
+
         if is_directed:
 
             interaction.add_evidence(
@@ -2703,31 +2736,14 @@ def remove_htp(self, threshold = 50, keep_directed = False):
             )
         )
 
-        interactions_per_reference = self.numof_interactions_per_reference()
-        interactions_by_reference = self.interactions_by_reference()
-
-        htp_refs = {
-            ref
-            for ref, cnt in iteritems(interactions_per_reference)
-            if cnt > threshold
-        }
-
-        to_remove = set()
+        to_remove = self.htp_interactions(
+            threshold = threshold,
+            ignore_directed = keep_directed,
+        )
 
         ecount_before = self.ecount
         vcount_before = self.vcount
 
-        for key, ia in iteritems(self.interactions):
-
-            if (
-                not ia.get_references() - htp_refs and (
-                    not keep_directed or
-                    not ia.is_directed()
-                )
-            ):
-
-                to_remove.add(key)
-
         for key in to_remove:
 
             self.remove_interaction(*key)
@@ -2746,6 +2762,54 @@ def remove_htp(self, threshold = 50, keep_directed = False):
         )
 
 
+    def htp_references(self, threshold = 50):
+        """
+        Collects the high-throughput references i.e. the ones cited at a
+        higher number of interactions than ``threshold``.
+        """
+
+        interactions_per_reference = self.numof_interactions_per_reference()
+
+        htp_refs = {
+            ref
+            for ref, cnt in iteritems(interactions_per_reference)
+            if cnt > threshold
+        }
+
+        self._log('High-throughput references collected: %u' % len(htp_refs))
+
+        return htp_refs
+
+
+    def htp_interactions(self, threshold = 50, ignore_directed = False):
+        """
+        Collects the interactions only from high-throughput studies.
+        
+        :returns:
+            Set of interaction keys (tuples of entities).
+        """
+
+        htp_refs = self.htp_references(threshold = threshold)
+
+        htp_int = set()
+
+        for key, ia in iteritems(self.interactions):
+
+            if (
+                (
+                    not ignore_directed or
+                    not ia.is_directed()
+                ) and
+                not ia.get_references() - htp_refs
+            ):
+
+                htp_int.add(key)
+
+        self._log('High-throughput interactions collected: %u' % len(htp_int))
+
+        return htp_int
+
+
     def remove_undirected(self, min_refs = None):
 
         self._log(
@@ -2918,6 +2982,7 @@ def load_transcription(
             exclude = None,
             reread = False,
             redownload = False,
+            allow_loops = None,
             **kwargs
         ):
 
@@ -2929,6 +2994,7 @@ def load_transcription(
                 levels = dorothea_levels,
                 reread = reread,
                 redownload = redownload,
+                allow_loops = allow_loops,
             )
 
         if original_resources:
@@ -2944,6 +3010,7 @@ def load_transcription(
                 reread = reread,
                 redownload = redownload,
                 exclude = exclude,
+                allow_loops = allow_loops,
             )
 
         if make_df:
@@ -2962,6 +3029,7 @@ def transcription(
             redownload = False,
             make_df = False,
             ncbi_tax_id = 9606,
+            allow_loops = None,
             **kwargs
         ):
         """
@@ -3824,6 +3892,32 @@ def _add_method(cls, method_name, method, signature = None, doc = None):
         )
 
 
+    def _allow_loops(self, allow_loops = None, resource = None):
+        """
+        Integrates settings for the `allow_loops` parameter from the
+        method, instance and module level settings.
+        """
+
+        default = settings.get('network_allow_loops')
+
+        return (
+            # from the arguments of the actual `load` call
+            allow_loops
+                if isinstance(allow_loops, bool) else
+            # from the current instance
+            self.allow_loops
+                if isinstance(self.allow_loops, bool) else
+            # interaction type specific settings from the module level
+            resource.networkinput.interaction_type in default
+                if (
+                    isinstance(default, common.list_like) and
+                    hasattr(resource, 'networkinput')
+                ) else
+            # general settings from the module level
+            bool(default)
+        )
+
+
 Network._generate_get_methods()
 Network._generate_partners_methods()
 Network._generate_count_methods()

src/pypath/omnipath/export.py

@@ -258,7 +258,15 @@ def process_interaction(self, ia):
             directed = bool(ia.direction[nodes])
             directed_rev = bool(ia.direction[tuple(reversed(nodes))])
 
-            if not directed and (_dir == 'b_a' or directed_rev):
+            if (
+                (
+                    not directed and
+                    (_dir == 'b_a' or directed_rev)
+                ) or (
+                    ia.is_loop() and
+                    _dir == 'b_a'
+                )
+            ):
 
                 continue
 

src/pypath/share/settings.py

@@ -84,6 +84,7 @@
     'deathdomain': 'deathdomain.tsv',
     'hpmr_preprocessed': 'hpmr_preprocessed.pickle',
     'network_expand_complexes': False,
+    'network_allow_loops': True,
     'network_keep_original_names': True,
     'network_pickle_cache': True,
     'go_pickle_cache': True,