Skip to content
This repository has been archived by the owner on Apr 12, 2023. It is now read-only.

Commit

Permalink
Merge pull request #103 from biothings/reasoner-output
Browse files Browse the repository at this point in the history 
Update Reasoner API Std output
  • Loading branch information
@kevinxin90
kevinxin90 authored May 12, 2020
2 parents 7f2acd1 + a6b9ce3 commit 3dde8c6
Show file tree
Hide file tree
Showing 2 changed files with 149 additions and 66 deletions.
102 changes: 62 additions & 40 deletions biothings_explorer/export/reasoner.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,9 +10,10 @@
from collections import defaultdict


class ReasonerConverter():
class ReasonerConverter:

"""Convert the output of BTE to ReasonerAPIStd."""

def __init__(self):
self.result = defaultdict(list)

Expand All @@ -23,8 +24,17 @@ def load_bte_query_path(self, start, intermediate, end):
:param: intermediate : the intermediate nodes connecting input and output
:param: end : the output of user query
"""
self.path = [start.get('type')]
self.path = [start.get("type")]
self.start = start
if ":" in self.start["primary"]["value"]:
self.start_node_curie = self.start["primary"]["value"]
else:
self.start_node_curie = (
self.start["primary"]["identifier"]
+ ":"
+ self.start["primary"]["value"]
)

if intermediate:
if isinstance(intermediate, list):
self.path += intermediate
Expand All @@ -35,7 +45,7 @@ def load_bte_query_path(self, start, intermediate, end):
elif isinstance(end, list):
self.path.append(tuple(end))
elif isinstance(end, dict):
self.path.append(end.get('type'))
self.path.append(end.get("type"))

def load_bte_output(self, G):
"""Load bte output in the format of networkx graph into class.
Expand All @@ -55,7 +65,11 @@ def get_curie(self, node):
node_info = self.nodes[node]
if "identifier" in node_info:
prefix = node_info["identifier"]
curie = prefix.upper() + ':' + node
# if the node id is already in curie format
# then no need to add prefix again
if ":" in node:
return node
curie = prefix + ":" + node
return curie
return node

Expand All @@ -74,40 +88,43 @@ def fetch_edges(self):
for k, v, o in self.G.edges(data=True):
source_id = self.get_curie(k)
target_id = self.get_curie(v)
edge_source = o['info'].get('$api')
_type = o.get('label')
edge_source = o["info"].get("$api")
_type = o.get("label")
_id = self.hash_id(source_id + target_id + edge_source + _type)
edge = {"source_id": source_id,
"target_id": target_id,
"edge_source": edge_source,
"id": _id,
"type": _type}
self.result[source_id + '|||' + target_id].append(_id)
edge = {
"source_id": source_id,
"target_id": target_id,
"edge_source": edge_source,
"id": _id,
"type": _type,
}
self.result[source_id + "|||" + target_id].append(_id)
edges.append(edge)
return edges

def fetch_nodes(self):
"""Reorganize the nodes into reasonerSTD format."""
nodes = []
for k, v in self.nodes:
name = v['equivalent_ids'].get("name")
name = v["equivalent_ids"].get("name")
if name and isinstance(name, list):
name = str(name[0])
else:
name = str(self.get_curie(k))
node = {"id": self.get_curie(k),
"name": name,
"type": v["type"],
"equivalent_identifiers": v['equivalent_ids']}
node = {
"id": self.get_curie(k),
"name": name,
"type": v["type"],
"equivalent_identifiers": v["equivalent_ids"],
}
nodes.append(node)
return nodes

def generate_knowledge_graph(self):
"""Reorganize the nodes and edges into reasonerSTD format."""
if len(self.G) == 0:
return {"nodes": [], "edges": []}
return {"nodes": self.fetch_nodes(),
"edges": self.fetch_edges()}
return {"nodes": self.fetch_nodes(), "edges": self.fetch_edges()}

def generate_question_graph(self):
if not self.path:
Expand All @@ -122,51 +139,56 @@ def generate_question_graph(self):

for i, node in enumerate(self.path):
if isinstance(node, str):
nodes.append({"id": "n" + str(node_id),
"type": node})
node2idmapping[str(i) + '-' + node + '-0'] = "n" + str(node_id)
nodes.append({"id": "n" + str(node_id), "type": node})
node2idmapping[str(i) + "-" + node + "-0"] = "n" + str(node_id)
node_id += 1
elif isinstance(node, tuple):
for j, _n in enumerate(node):
nodes.append({"id": "n" + str(node_id),
"type": _n})
node2idmapping[str(i) + '-' + _n + '-' + str(j)] = "n" + str(node_id)
nodes.append({"id": "n" + str(node_id), "type": _n})
node2idmapping[str(i) + "-" + _n + "-" + str(j)] = "n" + str(
node_id
)
node_id += 1
nodes[0]['curie'] = [self.start['primary']['identifier'] + ':' + self.start['primary']['value']]

nodes[0]["curie"] = self.start_node_curie

for i in range(0, len(self.path) - 1):
source_node = self.path[i]
target_node = self.path[i+1]
target_node = self.path[i + 1]
if isinstance(source_node, str):
source_node = [source_node]
if isinstance(target_node, str):
target_node = [target_node]
for p, _s in enumerate(source_node):
for q, _t in enumerate(target_node):
source_id = node2idmapping[str(i) + '-' + _s + '-' + str(p)]
target_id = node2idmapping[str(i+1) + '-' + _t + '-' + str(q)]
edges.append({"id": "e" + str(edge_id),
"source_id": source_id,
"target_id": target_id,
"directed": True})
source_id = node2idmapping[str(i) + "-" + _s + "-" + str(p)]
target_id = node2idmapping[str(i + 1) + "-" + _t + "-" + str(q)]
edges.append(
{
"id": "e" + str(edge_id),
"source_id": source_id,
"target_id": target_id,
"directed": True,
}
)
edge_id += 1
return {"edges": edges,
"nodes": nodes}
return {"edges": edges, "nodes": nodes}

def generate_result(self):
result = {"node_bindings": [], "edge_bindings": []}
if not self.result:
return result
result["node_bindings"].append({"qg_id": "n0", "kg_id": self.start_node_curie})
for k, v in self.result.items():
target_id = k.split('|||')[-1]
target_id = k.split("|||")[-1]
result["node_bindings"].append({"qg_id": "n1", "kg_id": target_id})
result["edge_bindings"].append({"qg_id": "e1", "kg_id": v})
return result

def generate_reasoner_response(self):
"""Generate reasoner response."""
response = {"query_graph": self.generate_question_graph(),
"knowledge_graph": self.generate_knowledge_graph()}
response['results'] = self.generate_result()
response = {
"query_graph": self.generate_question_graph(),
"knowledge_graph": self.generate_knowledge_graph(),
}
response["results"] = self.generate_result()
return response
113 changes: 87 additions & 26 deletions tests/test_bte2reasoner.py
View file
Original file line number Diff line number Diff line change
@@ -1,31 +1,92 @@
import unittest
from biothings_explorer.user_query_dispatcher import FindConnection
from biothings_explorer.user_query_dispatcher import (
FindConnection,
SingleEdgeQueryDispatcher,
)
from biothings_explorer.hint import Hint
import requests
import json

ht = Hint()
cxcr4 = ht.query('CXCR4')['Gene'][0]
fc = FindConnection(input_obj=cxcr4, output_obj='ChemicalSubstance', intermediate_nodes=None)
fc.connect(verbose=True)
response = fc.to_reasoner_std()

class TestSingleHopQuery(unittest.TestCase):

def test_result_section(self):
res = requests.post("http://transltr.io:7071/validate_result",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(response["results"])).json()
self.assertEqual(res, "Successfully validated")

def test_query_graph_section(self):
res = requests.post("http://transltr.io:7071/validate_querygraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(response["query_graph"])).json()
self.assertEqual(res, "Successfully validated")

def test_knowledge_graph_section(self):
res = requests.post("http://transltr.io:7071/validate_knowledgegraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(response["knowledge_graph"])).json()
self.assertEqual(res, "Successfully validated")

class TestFindConnection(unittest.TestCase):
@classmethod
def setUpClass(cls):
ht = Hint()
cxcr4 = ht.query("CXCR4")["Gene"][0]
fc = FindConnection(
input_obj=cxcr4, output_obj="ChemicalSubstance", intermediate_nodes=None
)
fc.connect(verbose=True)
cls.response = fc.to_reasoner_std()

def test_result_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_result",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["results"]),
).json()
cls.assertEqual(res, "Successfully validated")

def test_query_graph_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_querygraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["query_graph"]),
).json()
cls.assertEqual(res, "Successfully validated")

def test_knowledge_graph_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_knowledgegraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["knowledge_graph"]),
).json()
cls.assertEqual(res, "Successfully validated")


class TestSingleEdgeQuery(unittest.TestCase):
@classmethod
def setUpClass(cls):
seqd = SingleEdgeQueryDispatcher(
input_cls="ChemicalSubstance",
output_cls="Protein",
pred="",
input_id="CHEMBL.COMPOUND",
values="CHEMBL112",
)
seqd.query()
cls.response = seqd.to_reasoner_std()

def test_result_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_result",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["results"]),
).json()
cls.assertEqual(res, "Successfully validated")

def test_query_graph_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_querygraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["query_graph"]),
).json()
cls.assertEqual(res, "Successfully validated")

def test_knowledge_graph_section(cls):
res = requests.post(
"http://transltr.io:7071/validate_knowledgegraph",
headers={"accept": "text/plain", "content-type": "application/json"},
data=json.dumps(cls.response["knowledge_graph"]),
).json()
cls.assertEqual(res, "Successfully validated")

def test_knowledge_graph_node_id(cls):
nodes = [item["id"] for item in cls.response["knowledge_graph"]["nodes"]]
cls.assertIn("CHEMBL.COMPOUND:CHEMBL112", nodes)
cls.assertIn("PR:000011298", nodes)

def test_results_node_binding(cls):
nodes = [item["kg_id"] for item in cls.response["results"]["node_bindings"]]
cls.assertIn("CHEMBL.COMPOUND:CHEMBL112", nodes)
cls.assertIn("PR:000011298", nodes)

0 comments on commit 3dde8c6

Please sign in to comment.