From efe65a77b24bbc34a8e9e15a30c36583c836f494 Mon Sep 17 00:00:00 2001
From: AndiMajore <andi.majore@googlemail.com>
Date: Tue, 26 Jul 2022 00:09:44 +0200
Subject: [PATCH] adjusted id-space wise analysis preparation and execution
Former-commit-id: fb4d5146205053e00478b0de87642b09d692bc5a [formerly 45a041e729435626ada319eab340b8d962993623]
Former-commit-id: ed383169c0bf47eb77541dfc30c69e1894350858
---
drugstone/management/commands/make_graphs.py | 112 ++++++++++++++-----
drugstone/management/commands/populate_db.py | 28 ++---
drugstone/views.py | 83 +++++++-------
tasks/betweenness_centrality.py | 8 +-
tasks/closeness_centrality.py | 8 +-
tasks/degree_centrality.py | 8 +-
tasks/keypathwayminer_task.py | 17 ++-
tasks/multi_steiner.py | 7 +-
tasks/network_proximity.py | 4 +-
tasks/trust_rank.py | 2 +-
tasks/util/find_bridges.py | 2 +-
tasks/util/read_graph_tool_graph.py | 27 +----
tasks/util/scores_to_results.py | 22 ++--
tasks/util/steiner_tree.py | 2 +-
14 files changed, 192 insertions(+), 138 deletions(-)
diff --git a/drugstone/management/commands/make_graphs.py b/drugstone/management/commands/make_graphs.py
index 0ff9bb3..8794f48 100755
--- a/drugstone/management/commands/make_graphs.py
+++ b/drugstone/management/commands/make_graphs.py
@@ -1,3 +1,4 @@
+from collections import defaultdict
from typing import List, Tuple
import graph_tool.all as gt
from drugstone import models
@@ -77,7 +78,7 @@ def _internal_ppis(dataset) -> List[models.ProteinProteinInteraction]:
return node_node_interaction_objects
-def create_gt(params: Tuple) -> None:
+def create_gt(params: List[str]) -> None:
"""Fetches all required information to build a graph-tools file for given
PPI and PDI dataset names (params). Builds the graph-tools file and saves it in
the data/Networks folder.
@@ -85,37 +86,31 @@ def create_gt(params: Tuple) -> None:
Args:
params (Tuple[str, str]): Protein-protein-dataset name, Protein-drug-dataset name
"""
- ppi_dataset, pdi_dataset = params
+ ppi_dataset, pdi_dataset, identifier = params
+
licensed = ppi_dataset.licenced or pdi_dataset.licenced
# get data from api
g = gt.Graph(directed=False)
+
e_type = g.new_edge_property("string")
v_type = g.new_vertex_property("string")
v_name = g.new_vertex_property("string")
- v_drugstone_id = g.new_vertex_property("string")
- v_has_symbol = g.new_vertex_property("bool")
- v_has_entrez = g.new_vertex_property("bool")
- v_has_ensembl = g.new_vertex_property("bool")
- v_expression = g.new_vertex_property("string")
# for drugs
v_status = g.new_vertex_property("string")
v_drug_id = g.new_vertex_property("string")
+ v_internal_id = g.new_vertex_property("string")
g.edge_properties["type"] = e_type
- g.edge_properties["drugstone_id"] = e_type
+ # g.edge_properties["drugstone_id"] = e_type
g.vertex_properties["type"] = v_type
g.vertex_properties["name"] = v_name
- g.vertex_properties["drugstone_id"] = v_drugstone_id
- g.vertex_properties["has_symbol"] = v_has_symbol
- g.vertex_properties["has_entrez"] = v_has_entrez
- g.vertex_properties["has_ensembl"] = v_has_ensembl
g.vertex_properties["status"] = v_status
g.vertex_properties["drug_id"] = v_drug_id
- g.vertex_properties["expression"] = v_expression
+ g.vertex_properties["internal_id"] = v_internal_id
# store nodes to connect them when creating edges
vertices = {}
@@ -123,21 +118,46 @@ def create_gt(params: Tuple) -> None:
# add vertices
# print("adding nodes")
- print(f'loading nodes')
+ print(f'loading nodes for {identifier}')
# extend node data by cancer nodes, we create a normal node for each cancer node.
# on reading the data, we decide which one to keep based on the user selected cancer types
- has_ensembl_set = {node.protein_id for node in models.EnsemblGene.objects.all()}
+ is_entrez = identifier == 'entrez'
+ is_symbol = identifier == 'symbol'
+ is_uniprot = identifier == 'uniprot'
+ is_ensg = identifier == 'ensg'
+
+ if is_ensg:
+ ensembl_set = defaultdict(set)
+ for node in models.EnsemblGene.objects.all():
+ ensembl_set[node.protein_id].add(node.name)
+
+ node_id_map = defaultdict(set)
+ drugstone_ids_to_node_ids = defaultdict(set)
for node in models.Protein.objects.all():
+ if is_entrez:
+ if len(node.entrez) != 0:
+ node_id_map[node.entrez].add(node.id)
+ drugstone_ids_to_node_ids[node.id].add(node.entrez)
+ elif is_symbol:
+ if len(node.gene) != 0:
+ node_id_map[node.gene].add(node.id)
+ drugstone_ids_to_node_ids[node.id].add(node.gene)
+ elif is_uniprot:
+ node_id_map[node.uniprot_code].add(node.id)
+ drugstone_ids_to_node_ids[node.id].add(node.uniprot_code)
+ elif is_ensg:
+ for id in ensembl_set[node.id]:
+ node_id_map[id].add(node.id)
+ drugstone_ids_to_node_ids[node.id].add(id)
+
+ for id, nodes in node_id_map.items():
v = g.add_vertex()
v_type[v] = 'protein'
- v_drugstone_id[v] = f"p{node.id}"
- v_has_symbol[v] = len(node.gene) != 0
- v_has_entrez[v] = len(node.entrez) != 0
- v_has_ensembl[v] = node.id in has_ensembl_set
- vertices[node.id] = v
-
+ v_internal_id[v] = id
+ for drugstone_id in nodes:
+ vertices[drugstone_id] = v
print("done with nodes")
print(f"adding drugs")
@@ -145,22 +165,42 @@ def create_gt(params: Tuple) -> None:
v = g.add_vertex()
v_type[v] = 'drug'
v_status[v] = node.status
- v_drugstone_id[v] = f'dr{node.id}'
+ v_internal_id[v] = f'dr{node.id}'
drug_vertices[node.id] = v
+
print("done with drugs")
# add edges
print(f'adding ppi_edges/{ppi_dataset}')
+
+ uniq_edges = set()
+
for edge_raw in _internal_ppis(ppi_dataset):
- e = g.add_edge(vertices[edge_raw.from_protein_id], vertices[edge_raw.to_protein_id])
- e_type[e] = 'protein-protein'
+ id1 = edge_raw.from_protein_id
+ id2 = edge_raw.to_protein_id
+ if id1 > id2:
+ tmp = id1
+ id1 = id2
+ id2 = tmp
+ hash = f'{id1}_{id2}'
+ if hash not in uniq_edges and id1 in vertices and id2 in vertices:
+ uniq_edges.add(hash)
+ e = g.add_edge(vertices[id1], vertices[id2])
+ e_type[e] = 'protein-protein'
print("done with edges")
+ uniq_edges = set()
+
print(f'loading drug_edges/{pdi_dataset}')
for edge_raw in _internal_pdis(pdi_dataset):
- e = g.add_edge(drug_vertices[edge_raw.drug_id], vertices[edge_raw.protein_id])
- e_type[e] = 'drug-protein'
+ id1 = edge_raw.drug_id
+ id2 = edge_raw.protein_id
+ hash = f'{id1}_{id2}'
+ if hash not in uniq_edges and id1 in drug_vertices and id2 in vertices:
+ uniq_edges.add(hash)
+ e = g.add_edge(drug_vertices[id1], vertices[id2])
+ e_type[e] = 'drug-protein'
print("done with drug edges")
# remove unconnected proteins
@@ -177,7 +217,7 @@ def create_gt(params: Tuple) -> None:
g.remove_vertex(reversed(sorted(delete_vertices)), fast=True)
# save graph
- filename = f"./data/Networks/internal_{ppi_dataset.name}_{pdi_dataset.name}"
+ filename = f"./data/Networks/{identifier}_{ppi_dataset.name}-{pdi_dataset.name}"
if licensed:
filename += "_licenced"
filename += ".gt"
@@ -195,11 +235,25 @@ class Command(BaseCommand):
pdi_datasets = models.PDIDataset.objects.all()
+ licenced_ppi_dataset = {ppi.name: ppi for ppi in ppi_datasets if ppi.licenced}
+ licenced_pdi_dataset = {pdi.name: pdi for pdi in pdi_datasets if pdi.licenced}
+
+ uniq_combis = set()
parameter_combinations = []
for protein_interaction_dataset in ppi_datasets:
for pdi_dataset in pdi_datasets:
- parameter_combinations.append((protein_interaction_dataset, pdi_dataset))
-
+ licenced = protein_interaction_dataset.licenced or pdi_dataset.licenced
+ if licenced:
+ protein_interaction_dataset = licenced_ppi_dataset[
+ protein_interaction_dataset.name] if protein_interaction_dataset.name in licenced_ppi_dataset else protein_interaction_dataset
+ pdi_dataset = licenced_pdi_dataset[
+ pdi_dataset.name] if pdi_dataset.name in licenced_pdi_dataset else pdi_dataset
+ hash = f'{protein_interaction_dataset.name}-{pdi_dataset.name}_{licenced}'
+ if hash in uniq_combis:
+ continue
+ uniq_combis.add(hash)
+ for identifier in ['ensg', 'symbol', 'ensembl', 'uniprot']:
+ parameter_combinations.append([protein_interaction_dataset, pdi_dataset, identifier])
# close all database connections so subprocesses will create their own connections
# this prevents the processes from running into problems because of using the same connection
db.connections.close_all()
diff --git a/drugstone/management/commands/populate_db.py b/drugstone/management/commands/populate_db.py
index 429cf6f..1fd5f5f 100755
--- a/drugstone/management/commands/populate_db.py
+++ b/drugstone/management/commands/populate_db.py
@@ -215,20 +215,20 @@ def populate(kwargs):
print(f'Populated {n} DrDi associations from DrugBank.')
if kwargs['protein_protein']:
- print('Importing PPIs from unlicenced NeDRexDB...')
- n = NedrexImporter.import_protein_protein_interactions(importer,
- DatasetLoader.get_ppi_nedrex(nedrex_api_url_unlicenced, False),
- update)
- total_n += n
- print(f'Imported {n} PPIs from unlicended NeDRexDB')
- print('Importing PPIs from licenced NeDRexDB...')
- n = NedrexImporter.import_protein_protein_interactions(importer,
- DatasetLoader.get_ppi_nedrex(nedrex_api_url_licenced,
- True),
- update)
- total_n += n
- nedrex_update = True
- print(f'Imported {n} PPIs from licended NeDRexDB')
+ # print('Importing PPIs from unlicenced NeDRexDB...')
+ # n = NedrexImporter.import_protein_protein_interactions(importer,
+ # DatasetLoader.get_ppi_nedrex(nedrex_api_url_unlicenced, False),
+ # update)
+ # total_n += n
+ # print(f'Imported {n} PPIs from unlicended NeDRexDB')
+ # print('Importing PPIs from licenced NeDRexDB...')
+ # n = NedrexImporter.import_protein_protein_interactions(importer,
+ # DatasetLoader.get_ppi_nedrex(nedrex_api_url_licenced,
+ # True),
+ # update)
+ # total_n += n
+ # nedrex_update = True
+ # print(f'Imported {n} PPIs from licended NeDRexDB')
print('Populating PPIs from STRING...')
n = DataPopulator.populate_ppi_string(populator, DatasetLoader.get_ppi_string(), update)
total_n += n
diff --git a/drugstone/views.py b/drugstone/views.py
index d923139..d7a03fa 100755
--- a/drugstone/views.py
+++ b/drugstone/views.py
@@ -78,11 +78,9 @@ class TaskView(APIView):
licenced = parameters.get('licenced', False)
# find databases based on parameter strings
- print(get_ppi_ds(parameters.get('ppi_dataset', DEFAULTS['ppi']), licenced))
parameters['ppi_dataset'] = PPIDatasetSerializer().to_representation(
get_ppi_ds(parameters.get('ppi_dataset', DEFAULTS['ppi']), licenced))
- print(get_pdi_ds(parameters.get('pdi_dataset', DEFAULTS['pdi']), licenced))
parameters['pdi_dataset'] = PDIDatasetSerializer().to_representation(
get_pdi_ds(parameters.get('pdi_dataset', DEFAULTS['pdi']), licenced))
@@ -177,10 +175,6 @@ def map_nodes(request) -> Response:
nodes_mapped, id_key = query_proteins_by_identifier(node_ids, identifier)
# change data structure to dict in order to be quicker when merging
- # if identifier == 'ensg':
- # # a protein might have multiple ensg-numbers, unpack these into single nodes
- # nodes_mapped_dict = {node_id: node for node in nodes_mapped for node_id in node[id_key]}
- # else:
nodes_mapped_dict = {node[id_key][0]: node for node in nodes_mapped}
# merge fetched data with given data to avoid data loss
@@ -265,10 +259,14 @@ def result_view(request) -> Response:
drugs = []
network = result['network']
- node_types = {}
- node_attributes['node_types'] = node_types
- is_seed = {}
- node_attributes['is_seed'] = is_seed
+ node_types = node_attributes.get('node_types')
+ if not node_types:
+ node_types = {}
+ node_attributes['node_types'] = node_types
+ is_seed = node_attributes.get('is_seed')
+ if not is_seed:
+ is_seed = {}
+ node_attributes['is_seed'] = is_seed
scores = node_attributes.get('scores', {})
node_details = {}
protein_id_map = defaultdict(set)
@@ -286,7 +284,7 @@ def result_view(request) -> Response:
# merge input network with result network
for node in parameters['input_network']['nodes']:
# if node was already mapped, add user defined values to result of analysis
- if identifier in identifier_nodes:
+ if identifier in node:
node_name = node[identifier][0]
if node_name in node_details:
# update the node to not lose user input attributes
@@ -310,12 +308,32 @@ def result_view(request) -> Response:
result['node_attributes']['node_types'][node_id] = 'custom'
# extend the analysis network by the input netword nodes
# map edge endpoints to database proteins if possible and add edges to analysis network
-
+ protein_nodes = set()
# mapping all new protein and drug nodes by drugstoneIDs + adding scores
for node_id in nodes:
+ if node_id[:2] == 'dr':
+ node_data = DrugSerializer().to_representation(Drug.objects.get(id=int(node_id[2:])))
+ node_data['drugstoneType'] = 'drug'
+ drugs.append(node_data)
+ if node_id in scores:
+ node_data['score'] = scores.get(node_id, None)
+ node_types[node_id] = 'drug'
+ node_details[node_id] = node_data
+ elif node_id[:2] != 'di':
+ protein_nodes.add(node_id)
+ else:
+ continue
+
+ nodes_mapped, _ = query_proteins_by_identifier(protein_nodes, identifier)
+
+ nodes_mapped_dict = {node[identifier][0]: node for node in nodes_mapped}
- if node_id[0] == 'p':
- node_data = ProteinNodeSerializer().to_representation(Protein.objects.get(id=int(node_id[1:])))
+ # merge fetched data with given data to avoid data loss
+ for node_id in nodes:
+ if node_id in nodes_mapped_dict:
+ # node.update(nodes_mapped_dict[node['id']])
+ node_data = nodes_mapped_dict[node_id]
+ node_data['drugstoneType'] = 'protein'
# proteins.append(node_data)
node_ident = node_data[identifier][0]
# node_data[identifier] = [node_ident]
@@ -326,36 +344,20 @@ def result_view(request) -> Response:
score = scores.get(node_id, None)
if node_ident in node_details:
data = node_details[node_ident]
- data['entrez'].extend(node_data['entrez'])
- data['ensg'].extend(node_data['ensg'])
- data['symbol'].extend(node_data['symbol'])
- data['uniprot_ac'].extend(node_data['uniprot_ac'])
- if score:
- if 'score' in data:
- data['score'].append(score)
- else:
- data['score'] = [score] if score else []
+ data['score'] = [score] if score else None
else:
- node_data['score'] = [score] if score else []
+ node_data['score'] = score if score else None
node_data['drugstoneType'] = 'protein'
node_data['id'] = node_ident
node_data['label'] = node_ident
node_details[node_ident] = node_data
- elif node_id[:2] == 'dr':
- node_data = DrugSerializer().to_representation(Drug.objects.get(id=int(node_id[2:])))
- drugs.append(node_data)
- if node_id in scores:
- node_data['score'] = scores.get(node_id, None)
- node_types[node_id] = 'drug'
- node_details[node_id] = node_data
- else:
- continue
for node_id, detail in node_details.items():
- detail['symbol'] = list(set(detail['symbol']))
- detail['entrez'] = list(set(detail['entrez']))
- detail['uniprot_ac'] = list(set(detail['uniprot_ac']))
- detail['ensg'] = list(set(detail['ensg']))
+ if 'drugstoneType' in detail and detail['drugstoneType'] == 'protein':
+ detail['symbol'] = list(set(detail['symbol']))
+ detail['entrez'] = list(set(detail['entrez']))
+ detail['uniprot_ac'] = list(set(detail['uniprot_ac']))
+ detail['ensg'] = list(set(detail['ensg']))
edges = parameters['input_network']['edges']
edge_endpoint_ids = set()
@@ -389,7 +391,12 @@ def result_view(request) -> Response:
map(lambda n: {"from": f'p{n.from_protein_id}', "to": f'p{n.to_protein_id}'}, interaction_objects))
edges.extend(auto_edges)
result['network']['edges'].extend(edges)
- result['network']['nodes'] = list(identifier_nodes)
+ # uniq_edges = dict()
+ # for edge in result['network']['edges']:
+ # hash = edge['from'] + edge['to']
+ # uniq_edges[hash] = edge
+ # result['network']['edges']=list(uniq_edges.values())
+ # result['network']['nodes'] = list(identifier_nodes)
if 'scores' in result['node_attributes']:
del result['node_attributes']['scores']
diff --git a/tasks/betweenness_centrality.py b/tasks/betweenness_centrality.py
index f763b1e..06d4ca0 100755
--- a/tasks/betweenness_centrality.py
+++ b/tasks/betweenness_centrality.py
@@ -172,15 +172,13 @@ def betweenness_centrality(task_hook: TaskHook):
id_space = task_hook.parameters["config"].get("identifier","symbol")
- print(id_space)
-
# Parsing input file.
task_hook.set_progress(0 / 3.0, "Parsing input.")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
if ppi_dataset['licenced'] or pdi_dataset['licenced']:
filename += "_licenced"
- filename = os.path.join(task_hook.data_directory, filename+".gt")
- g, seed_ids, id_space, drug_ids = read_graph_tool_graph(
+ filename = os.path.join(task_hook.data_directory, filename + ".gt")
+ g, seed_ids, drug_ids = read_graph_tool_graph(
filename,
seeds,
max_deg,
diff --git a/tasks/closeness_centrality.py b/tasks/closeness_centrality.py
index 78222aa..7ae2772 100755
--- a/tasks/closeness_centrality.py
+++ b/tasks/closeness_centrality.py
@@ -170,13 +170,13 @@ def closeness_centrality(task_hook: TaskHook):
# Parsing input file.
task_hook.set_progress(0 / 4.0, "Parsing input.")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
- if ppi_dataset['licenced'] or pdi_dataset['licenced']:
- filename += "_licenced"
id_space = task_hook.parameters["config"].get("identifier", "symbol")
- filename = os.path.join(task_hook.data_directory, filename+".gt")
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
+ if ppi_dataset['licenced'] or pdi_dataset['licenced']:
+ filename += "_licenced"
+ filename = os.path.join(task_hook.data_directory, filename + ".gt")
# g, seed_ids, viral_protein_ids, drug_ids = read_graph_tool_graph(file_path, seeds, datasets, ignored_edge_types, max_deg, ignore_non_seed_baits, include_indirect_drugs, include_non_approved_drugs)
g, seed_ids, drug_ids = read_graph_tool_graph(filename, seeds, id_space, max_deg, include_indirect_drugs, include_non_approved_drugs, search_target)
task_hook.set_progress(1 / 4.0, "Computing edge weights.")
diff --git a/tasks/degree_centrality.py b/tasks/degree_centrality.py
index 1a709a0..e529c8a 100755
--- a/tasks/degree_centrality.py
+++ b/tasks/degree_centrality.py
@@ -150,13 +150,13 @@ def degree_centrality(task_hook: TaskHook):
# Parsing input file.
task_hook.set_progress(0 / 3.0, "Parsing input.")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
- if ppi_dataset['licenced'] or pdi_dataset['licenced']:
- filename += "_licenced"
id_space = task_hook.parameters["config"].get("identifier", "symbol")
- filename = os.path.join(task_hook.data_directory, filename+".gt")
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
+ if ppi_dataset['licenced'] or pdi_dataset['licenced']:
+ filename += "_licenced"
+ filename = os.path.join(task_hook.data_directory, filename + ".gt")
# g, seed_ids, viral_protein_ids, drug_ids = read_graph_tool_graph(file_path, seeds, datasets, ignored_edge_types, max_deg, ignore_non_seed_baits, False, include_non_approved_drugs)
g, seed_ids, drug_ids = read_graph_tool_graph(filename, seeds, id_space, max_deg, False, include_non_approved_drugs, search_target)
diff --git a/tasks/keypathwayminer_task.py b/tasks/keypathwayminer_task.py
index dc26914..54080de 100755
--- a/tasks/keypathwayminer_task.py
+++ b/tasks/keypathwayminer_task.py
@@ -10,7 +10,7 @@ import requests
from tasks.task_hook import TaskHook
-from drugstone.models import Protein
+from drugstone.models import Protein, EnsemblGene
# Base URL
# url = 'http://172.25.0.1:9003/keypathwayminer/requests/'
@@ -57,9 +57,18 @@ def kpm_task(task_hook: TaskHook):
# --- Fetch and generate the datasets
dataset_name = 'indicatorMatrix'
indicator_matrix_string = ''
- protein_backend_ids = [int(seed[1:]) for seed in task_hook.seeds]
- proteins = Protein.objects.filter(id__in=protein_backend_ids)
-
+ id_space = task_hook.parameters["config"].get("identifier", "symbol")
+ proteins = []
+ if id_space == 'symbol':
+ proteins = Protein.objects.filter(gene__in=task_hook.seeds)
+ elif id_space == 'entrez':
+ proteins = Protein.objects.filter(entrez__in=task_hook.seeds)
+ elif id_space == 'uniprot':
+ proteins = Protein.objects.filter(uniprot_code__in=task_hook.seeds)
+ elif id_space == 'ensg':
+ protein_ids = {ensg.protein_id for ensg in EnsemblGene.objects.filter(name__in=task_hook.seeds)}
+ proteins = Protein.objects.filter(id__in=protein_ids)
+ protein_backend_ids = {p.id for p in proteins}
for protein in proteins:
indicator_matrix_string += f'{protein.uniprot_code}\t1\n'
diff --git a/tasks/multi_steiner.py b/tasks/multi_steiner.py
index 6e81bca..f361168 100755
--- a/tasks/multi_steiner.py
+++ b/tasks/multi_steiner.py
@@ -97,7 +97,7 @@ def multi_steiner(task_hook: TaskHook):
search_target = task_hook.parameters.get("target", "drug-target")
- node_name_attribute = "drugstone_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
+ node_name_attribute = "internal_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
# Set number of threads if OpenMP support is enabled.
if gt.openmp_enabled():
@@ -108,11 +108,10 @@ def multi_steiner(task_hook: TaskHook):
id_space = task_hook.parameters["config"].get("identifier", "symbol")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
if ppi_dataset['licenced'] or pdi_dataset['licenced']:
filename += "_licenced"
-
- filename = os.path.join(task_hook.data_directory, filename+".gt")
+ filename = os.path.join(task_hook.data_directory, filename + ".gt")
g, seed_ids, _ = read_graph_tool_graph(filename, seeds, id_space, max_deg, target=search_target)
# seed_map = {g.vertex_properties["name"][node]: node for node in seed_ids}
seed_map = {g.vertex_properties[node_name_attribute][node]: node for node in seed_ids}
diff --git a/tasks/network_proximity.py b/tasks/network_proximity.py
index 6755d8e..716c7a3 100755
--- a/tasks/network_proximity.py
+++ b/tasks/network_proximity.py
@@ -89,10 +89,10 @@ def network_proximity(task_hook: TaskHook):
id_space = task_hook.parameters["config"].get("identifier", "symbol")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
if ppi_dataset['licenced'] or pdi_dataset['licenced']:
filename += "_licenced"
- filename = os.path.join(task_hook.data_directory, filename+".gt")
+ filename = os.path.join(task_hook.data_directory, filename + ".gt")
# g, seed_ids, _, drug_ids = read_graph_tool_graph(file_path, seeds, "", "", max_deg, False, True, include_non_approved_drugs)
g, seed_ids, drug_ids = read_graph_tool_graph(filename, seeds, id_space, max_deg, True, include_non_approved_drugs, target=search_target)
# Computing edge weights.
diff --git a/tasks/trust_rank.py b/tasks/trust_rank.py
index 4737bbe..fbcb5cc 100755
--- a/tasks/trust_rank.py
+++ b/tasks/trust_rank.py
@@ -201,7 +201,7 @@ def trust_rank(task_hook: TaskHook):
id_space = task_hook.parameters["config"].get("identifier", "symbol")
- filename = f"internal_{ppi_dataset['name']}_{pdi_dataset['name']}"
+ filename = f"{id_space}_{ppi_dataset['name']}-{pdi_dataset['name']}"
if ppi_dataset['licenced'] or pdi_dataset['licenced']:
filename += "_licenced"
filename = os.path.join(task_hook.data_directory, filename+".gt")
diff --git a/tasks/util/find_bridges.py b/tasks/util/find_bridges.py
index 890c4f5..fb8cc98 100755
--- a/tasks/util/find_bridges.py
+++ b/tasks/util/find_bridges.py
@@ -17,7 +17,7 @@ def __dfs_find_bridges(g, node, visited, disc, low, parent, is_bridge):
low[node] = min(low[node], low[nb])
if low[nb] > disc[node]:
is_bridge[g.edge(node, nb)] = True
- elif int(nb) != parent[node]: #TODO can in theory be removed because
+ elif int(nb) != parent[node]: #TODO can in theory be removed
low[node] = min(low[node], disc[nb])
def find_bridges(g):
diff --git a/tasks/util/read_graph_tool_graph.py b/tasks/util/read_graph_tool_graph.py
index d496f3b..92c1333 100755
--- a/tasks/util/read_graph_tool_graph.py
+++ b/tasks/util/read_graph_tool_graph.py
@@ -3,7 +3,8 @@ import graph_tool.topology as gtt
# def read_graph_tool_graph(file_path, seeds, datasets, ignored_edge_types, max_deg, ignore_non_seed_baits=False, include_indirect_drugs=False, include_non_approved_drugs=False):
-def read_graph_tool_graph(file_path, seeds,id_space, max_deg, include_indirect_drugs=False, include_non_approved_drugs=False,
+def read_graph_tool_graph(file_path, seeds, id_space, max_deg, include_indirect_drugs=False,
+ include_non_approved_drugs=False,
target='drug'):
r"""Reads a graph-tool graph from file.
@@ -45,7 +46,7 @@ def read_graph_tool_graph(file_path, seeds,id_space, max_deg, include_indirect_d
# drug_protein = "DrugHasTarget"
d_type = "drug"
- node_name_attribute = "drugstone_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
+ node_name_attribute = "internal_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
# Delete all nodes that are not contained in the selected datasets and have degrees higher than max_deg
deleted_nodes = []
for node in range(g.num_vertices()):
@@ -59,20 +60,18 @@ def read_graph_tool_graph(file_path, seeds,id_space, max_deg, include_indirect_d
# remove all drugs from graph if we are not looking for drugs
elif target != 'drug' and g.vertex_properties["type"][node] == d_type:
deleted_nodes.append(node)
- g.remove_vertex(deleted_nodes, fast=True)
+
+ g.remove_vertex(reversed(sorted(deleted_nodes)), fast=True)
# Retrieve internal IDs of seed_ids
seeds = set(seeds)
- print(seeds)
seed_ids = {}
drug_ids = []
- # is_matched = {protein: False for protein in seeds}
for node in range(g.num_vertices()):
node_type = g.vertex_properties["type"][node]
seed_id = g.vertex_properties[node_name_attribute][node]
if seed_id in seeds:
seed_ids[node] = seed_id
- # is_matched[seed_id] = node
if node_type == d_type:
if include_non_approved_drugs:
drug_ids.append(node)
@@ -81,16 +80,6 @@ def read_graph_tool_graph(file_path, seeds,id_space, max_deg, include_indirect_d
if "approved" in drug_groups:
drug_ids.append(node)
- # Check that all seed seeds have been matched and throw error, otherwise.
- # print(deleted_nodes)
- # print(seed_ids)
- # seeds = set(seed_ids.values())
- # for (node, seed_id) in seed_ids.items():
- # if is_matched[node]
- # for protein, found in is_matched.items():
- # if not found:
- # raise ValueError("Invalid seed protein {}. No node named {} in {}.".format(protein, protein, file_path))
-
# Delete edges that should be ignored or are not contained in the selected dataset.
deleted_edges = []
@@ -138,17 +127,11 @@ def read_graph_tool_graph(file_path, seeds,id_space, max_deg, include_indirect_d
for edge in deleted_edges:
g.remove_edge(edge)
g.set_fast_edge_removal(fast=False)
- print("Drugs")
- print(drug_ids)
- print("Vertices")
vertices = 0
for _ in g.vertices():
vertices += 1
- print(f'\t{vertices}')
- print("Edges")
edges = 0
for _ in g.edges():
edges += 1
- print(f'\t{edges}')
# Return the graph and the indices of the seed_ids and the seeds.
return g, list(seed_ids.keys()), drug_ids
diff --git a/tasks/util/scores_to_results.py b/tasks/util/scores_to_results.py
index 4f47663..e3b679e 100755
--- a/tasks/util/scores_to_results.py
+++ b/tasks/util/scores_to_results.py
@@ -15,7 +15,7 @@ def scores_to_results(
r"""Transforms the scores to the required result format."""
- node_name_attribute = "drugstone_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
+ node_name_attribute = "internal_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
candidates = []
# if strain_or_drugs == "drugs":
if target == "drug":
@@ -23,8 +23,6 @@ def scores_to_results(
else:
candidates = [(node, scores[node]) for node in range(g.num_vertices()) if scores[node] > 0 and node not in set(seed_ids)]
best_candidates = [item[0] for item in sorted(candidates, key=lambda item: item[1], reverse=True)[:result_size]]
- print(f'Candidate list length: {len(best_candidates)}')
-
# Concatenate best result candidates with seeds and compute induced subgraph.
# since the result size filters out nodes, the result network is not complete anymore.
# Therefore, it is necessary to find the shortest paths to the found nodes in case intermediate nodes have been removed.
@@ -36,6 +34,7 @@ def scores_to_results(
returned_nodes = set(seed_ids) # return seed_ids in any case
# return only the path to a drug with the shortest distance
+ accepted_candidates = set()
if filterPaths:
for candidate in best_candidates:
distances = gtt.shortest_distance(g, candidate, seed_ids)
@@ -53,11 +52,12 @@ def scores_to_results(
break
if drug_in_path:
continue
-
+ accepted_candidates.add(g.vertex_properties[node_name_attribute][int(candidate)])
for vertex in vertices:
if int(vertex) not in returned_nodes:
# inserting intermediate node in order to make result comprehensive
- intermediate_nodes.add(g.vertex_properties[node_name_attribute][int(vertex)])
+ if vertex != candidate:
+ intermediate_nodes.add(g.vertex_properties[node_name_attribute][int(vertex)])
returned_nodes.add(int(vertex))
for edge in edges:
if ((edge.source(), edge.target()) not in returned_edges) or ((edge.target(), edge.source()) not in returned_edges):
@@ -74,18 +74,21 @@ def scores_to_results(
break
if drug_in_path:
continue
-
+ accepted_candidates.add(g.vertex_properties[node_name_attribute][int(candidate)])
for vertex in vertices:
if int(vertex) not in returned_nodes:
# inserting intermediate node in order to make result comprehensive
- intermediate_nodes.add(g.vertex_properties[node_name_attribute][int(vertex)])
+ if vertex != candidate:
+ intermediate_nodes.add(g.vertex_properties[node_name_attribute][int(vertex)])
returned_nodes.add(int(vertex))
for edge in edges:
if ((edge.source(), edge.target()) not in returned_edges) or ((edge.target(), edge.source()) not in returned_edges):
returned_edges.add((edge.source(), edge.target()))
- print(f'Returned nodes number: {len(returned_nodes)}')
+ for node in accepted_candidates:
+ if node in intermediate_nodes:
+ intermediate_nodes.remove(node)
subgraph = {
- "nodes": [g.vertex_properties[node_name_attribute][node] for node in returned_nodes],
+ "nodes":[g.vertex_properties[node_name_attribute][node] for node in returned_nodes],
"edges": [{"from": g.vertex_properties[node_name_attribute][source], "to": g.vertex_properties[node_name_attribute][target]} for source, target in returned_edges],
}
@@ -97,6 +100,7 @@ def scores_to_results(
return {
"network": subgraph,
'intermediate_nodes': list(intermediate_nodes),
+ 'target_nodes': list(accepted_candidates),
"node_attributes":
{
"node_types": node_types,
diff --git a/tasks/util/steiner_tree.py b/tasks/util/steiner_tree.py
index 27cbedb..f91f647 100755
--- a/tasks/util/steiner_tree.py
+++ b/tasks/util/steiner_tree.py
@@ -5,7 +5,7 @@ import itertools as it
def steiner_tree(g, seeds, seed_map, weights, non_zero_hub_penalty):
- node_name_attribute = "drugstone_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
+ node_name_attribute = "internal_id" # nodes in the input network which is created from RepoTrialDB have primaryDomainId as name attribute
mc = gt.Graph(directed=False)
eprop_dist = mc.new_edge_property("int")
mc.ep['dist'] = eprop_dist
--
GitLab