5 |
6 | ## Materials and Methods
7 | Our ultimate goal was to build a biomedical knowledge graph via comprehensively incorporating biomedical knowledge as much as possible. To this end, we collected and integrated 17 publicly available data sources to curate a comprehensive one. Details of the used data resources were listed in [Table](https://github.com/wcm-wanglab/iBKH/blob/main/Source%20Information/README.md).
8 |
9 | ## Statistics of iBKH
10 | | Entity Type | Number | Included Identifiers |
11 | | ---------------|:---------:|:--------------------:|
12 | | Anatomy | 23,003 | Uberon ID, BTO ID, MeSH ID, Cell Ontology ID |
13 | | Disease | 19,236 | Disease Ontology ID, PharmGKB ID, MeSH ID, OMIM ID |
14 | | Drug | 37,997 | DrugBank ID, PharmGKB ID, MeSH ID |
15 | | Gene | 88,376 | HGNC ID, NCBI ID, PharmGKB ID |
16 | | Molecule | 2,065,015 | CHEMBL ID, CHEBI ID |
17 | | Symptom | 1,361 | MeSH ID |
18 | | Dietary Supplement Ingredient | 4,101 | iDISK ID |
19 | | Dietary Supplement Product | 137,568 | iDISK ID |
20 | | Therapeutic Class | 605 | iDISK ID, UMLS CUI |
21 | | Pathway | 2,988 | Reactome ID, Gene Ontology ID |
22 | | Side-Effect | 4,251 | UMLS CUI |
23 | | **Total Entities** | **2,384,501** | - |
24 |
25 | | Relation Type | Number |
26 | | ----------------|:----------:|
27 | | Anatomy-Gene | 12,171,021 |
28 | | Drug-Disease | 2,717,947 |
29 | | Drug-Drug | 2,684,682 |
30 | | Drug-Gene | 1,303,747 |
31 | | Disease-Disease | 11,072 |
32 | | Disease-Gene | 27,538,774 |
33 | | Disease-Symptom | 3,357 |
34 | | Gene-Gene | 735,156 |
35 | | DSI-Symptom | 2,093 |
36 | | DSI-Disease | 5,134 |
37 | | DSI-Drug | 3,057 |
38 | | DSI-Anatomy | 4,334 |
39 | | DSP-DSI | 689,297 |
40 | | DSI-TC | 5,430 |
41 | | Disease-Pathway | 1,941 |
42 | | Drug-Pathway | 3,231 |
43 | | Gene-Pathway | 152,243 |
44 | | Drug-Side Effect| 163,206 |
45 | | **Total Relations** | **48,194,646** |
46 |
47 | ## Neo4j Deployment
48 | We deployed our iBKH using Neo4j (https://neo4j.com) on AWS, a robust graph database platform. Specifically, Neo4j can take the CSV files of entities and relations we curated as input and automatically create a KG instance. In this way, the iBKH can be updated efficiently and flexibly. Follow instructions [here](https://docs.google.com/document/d/1cLDPLp_nVCJ5xrDlJ-B-Q3wf24tb-Dyq55nAXxaNgTM/edit?usp=sharing) to deploy iBKH in your AWS server.
49 |
50 | ## iBKH-based knowledge discovery
51 | We developed a knowledge discovery pipeline in iBKH. We utilized [Deep Graph Library - Knowledge Graph Embedding (DGL-KE)](https://github.com/awslabs/dgl-ke) to learn embeddings of iBKH, based on which we can derive novel biomedical knowledge. We applied our pipeline for two case studies, Alzheimer's disease drug repurposing hypothesis generation and a knowledge-enhanced EHR patient cohort study. All codes for these analyses can be found [here](https://github.com/wcm-wanglab/iBKH/tree/main/Codes). And please refer to [readme for case studies](https://github.com/wcm-wanglab/iBKH/blob/main/Codes/README.md) for more details.
52 |
53 | ## Licence
54 | iBKH is licensed under the [Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0) and [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/). The iBKH integrated the data from many resources, and users should consider the licenses for each of them (see the detail in the [table](https://github.com/wcm-wanglab/iBKH/blob/main/Source%20Information/README.md)). For sources with defined licenses, we apply the license attribute on a per-node and per-edge basis. However, some sources do not provide any licenses, so for those, we have requested permission.
55 |
56 | ## Cite
57 | ```
58 | @article {Su2021.03.12.21253461,
59 | title = {Biomedical Discovery through the integrative Biomedical Knowledge Hub (iBKH)},
60 | author = {Chang Su, Yu Hou, Suraj Rajendran, Jacqueline R. M. A. Maasch, Zehra Abedi, Haotan Zhang, Zilong Bai,
61 | Anthony Cuturrufo, Winston Guo, Fayzan F. Chaudhry, Gregory Ghahramani, Jian Tang, Feixiong Cheng,
62 | Yue Li, Rui Zhang, Jiang Bian, Fei Wang},
63 | year = {2022},
64 | doi = {10.1101/2021.03.12.21253461},
65 | publisher = {Cold Spring Harbor Laboratory Press},
66 | URL = {https://www.medrxiv.org/content/10.1101/2021.03.12.21253461v4},
67 | journal = {medRxiv}
68 | }
69 |
70 | ```
71 |
72 |
--------------------------------------------------------------------------------
/iBKH-KD-protocol/funcs/knowledge_visualization.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python3
2 | # -*- coding: utf-8 -*-
3 | """
4 | Created on Wed Jul 26 17:48:29 2023
5 |
6 | @author: changsu
7 | """
8 |
9 |
10 | import networkx as nx
11 | from neo4j import GraphDatabase
12 | import matplotlib.pyplot as plt
13 | import os
14 |
15 |
16 |
17 |
18 |
19 | def subgraph_visualization(target_type, target_list, predicted_type, predicted_list,
20 | excluded_r_type = [],
21 | neo4j_url = "neo4j://54.210.251.104:7687",
22 | username = "neo4j", password = "password",
23 | alpha = 1, k=0.3, nsize=200, target_size_ratio=2.5,
24 | with_node_label=True, node_label_size = 10,
25 | with_edge_label=True, edge_label_size = 7,
26 | figsize=(14, 10),
27 | save=True, save_path='output', save_name=None):
28 |
29 | # Connect to the Neo4j database
30 | driver = GraphDatabase.driver(neo4j_url,
31 | auth=(username, password),
32 | encrypted=False)
33 | neo4j_res_list = []
34 |
35 | # Build Cypher statement
36 | for target in target_list:
37 | for predict in predicted_list:
38 | cypher = "MATCH (e1:" + target_type + " {Name: \"" + target + "\"})"
39 | cypher += ", (e2:" + predicted_type + " {Name: \"" + predict + "\"})"
40 | cypher += ", p = allShortestPaths((e1)-[*..5]-(e2)) RETURN p LIMIT 30"
41 |
42 | # Run the Neo4j query and retrieve the results
43 | session = driver.session()
44 | neo4j_res = session.run(cypher)
45 | neo4j_res_list.append(neo4j_res)
46 |
47 | # Create a NetworX Graph object
48 | g = nx.MultiGraph()
49 |
50 | # Define node groups and their corresponding colors
51 | group_colors = {
52 | "Disease": "#E0C3FC",
53 | "Drug": "#83B5D1",
54 | "Gene": "#F28482",
55 | "Symptom": "#7B967A",
56 | "Side-effect": "#9DA1DD",
57 | "Pathway": "#94D2BD"
58 | }
59 |
60 |
61 | node_id_map = {}
62 | id_node_map = {}
63 | node_color = {}
64 |
65 | edge_label_map = {}
66 | # Iterate over the Neo4j query result and add nodes to the network
67 | idx = 0
68 | for neo4j_res in neo4j_res_list:
69 | for record in neo4j_res:
70 | path = record["p"]
71 |
72 | # adding node
73 | for node in path.nodes:
74 | node_type_list = list(node.labels)
75 | node_type = node_type_list[0] if node_type_list[0] != 'Node' else node_type_list[1]
76 | node_name = node['Name']
77 |
78 | if node_name not in node_id_map:
79 | node_id_map[node_name] = idx
80 | id_node_map[idx] = node_name
81 |
82 | g.add_node(node_name)
83 | node_color[node_name] = group_colors[node_type]
84 |
85 | idx += 1
86 |
87 | # adding edges
88 | for relation in path.relationships:
89 | start_node_type = list(relation.start_node.labels)[0]
90 | end_node_type = list(relation.end_node.labels)[0]
91 |
92 | r_type = relation.type
93 |
94 | if r_type in excluded_r_type:
95 | continue
96 |
97 | start = relation.start_node['Name']
98 | end = relation.end_node["Name"]
99 |
100 | edge_label_map[(start, end)] = r_type
101 |
102 | g.add_edge(start, end, label=r_type)
103 |
104 |
105 | color_map = []
106 | size_map = []
107 | for n in g.nodes():
108 | color_map.append(node_color[n])
109 | if (n in target_list) or (n in predicted_list):
110 | size_map.append(nsize * target_size_ratio)
111 | else:
112 | size_map.append(nsize)
113 |
114 |
115 | plt.figure(figsize=figsize)
116 |
117 | positions = nx.spring_layout(g, k=k)
118 |
119 |
120 | for node in target_list:
121 | positions[node][0] -= alpha # adjust value as needed
122 |
123 | for node in predicted_list:
124 | positions[node][0] += alpha # adjust value as needed
125 |
126 | nx.draw(g, with_labels = with_node_label,
127 | node_color=color_map, node_size=size_map,
128 | edge_color='#E0E0E0', pos=positions,
129 | font_size=node_label_size)
130 |
131 | nx.draw_networkx_edge_labels(
132 | g, positions,
133 | edge_labels=edge_label_map,
134 | font_color='blue',
135 | font_size = edge_label_size
136 | )
137 |
138 | if save == True:
139 |
140 | if not os.path.exists(save_path):
141 | os.makedirs(save_path)
142 |
143 | if save_name == None:
144 | save_name ='shortest_path_interpretation_%s_%s.pdf' % (target_type, predicted_type)
145 | plt.savefig(save_path + '/' + save_name)
146 |
147 |
148 | plt.show()
149 |
150 | return g
151 |
152 |
153 | #excluded_r_type = [
154 | # 'Inferred_Relation_DDi', 'Semantic_Relation_DDi'
155 | # 'Semantic_Relation_DG', '19 Semantic_Relation_DiG',
156 | # 'Semantic_Relation_GG', 'Inferred_Relation_DiG']
157 |
158 |
--------------------------------------------------------------------------------
/Source Information/README.md:
--------------------------------------------------------------------------------
1 | **Source information**
2 |
3 | | Source | 7 |Entity | 8 |Relation | 9 |URL | 10 |License | 11 |||
|---|---|---|---|---|---|---|
| Type | 14 |Number | 15 |Type | 16 |Number | 17 ||||
| Bgee | 22 |Anatomy, Gene | 23 |60,072 | 24 |Anatomy-Express Present-Gene, Anatomy-Express Absent-Gene | 25 |11,731,369 | 26 |https://bgee.org/ | 27 |https://creativecommons.org/publicdomain/zero/1.0/ | 28 |
| Brenda Tissue Ontology | 31 |Tissue | 32 |6,478 | 33 |- | 34 |- | 35 |https://www.brenda-enzymes.org/index.php | 36 |https://creativecommons.org/licenses/by/4.0/ | 37 |
| Cell Ontology | 40 |Cell | 41 |2,200 | 42 |- | 43 |- | 44 |http://obofoundry.org/ontology/cl.html | 45 |https://creativecommons.org/licenses/by/4.0/ | 46 |
| Comparative Toxicogenomics Database | 49 |Disease, Gene, Chemical, Pathway | 50 |73,922 | 51 |Chemical-Gene, Chemical-Disease, Chemical-Pathway, Gene-Disease, Gene-Pathway, Disease-Pathway | 52 |38,344,568 | 53 |http://ctdbase.org/ | 54 |Confirmed via email | 55 |
| ChEMBL | 58 |Molecular | 59 |1,940,733 | 60 |- | 61 |- | 62 |https://www.ebi.ac.uk/chembl/ | 63 |https://creativecommons.org/licenses/by-sa/3.0/ | 64 |
| ChEBI | 67 |Molecular | 68 |155,342 | 69 |- | 70 |- | 71 |https://www.ebi.ac.uk/chebi/init.do | 72 |https://creativecommons.org/licenses/by/4.0/ | 73 |
| Drug Repurposing Knowledge Graph | 76 |Anatomy, Atc, Biological process, Cellular component, Compound, Disease, Gene, Molecular function, Pathway, Pharmacologic class, Side effect, Symptom, Tax | 77 |97,238 | 78 |Gene-Gene, Compound-Gene, Disease-Gene, Atc-Compound, Compound-Compound, Compound-Disease, Gene-Tax, Biological process-Gene, Disease-Symptom, Anatomy-Disease, Disease-Disease, Anatomy-Gene, Gene-Molecular function, Compound-Pharmacologic class, Cellular component-Gene, Gene-Pathway, Compound-Side effect | 79 |5,874,261 | 80 |https://github.com/gnn4dr/DRKG | 81 |https://www.apache.org/licenses/LICENSE-2.0 | 82 |
| Disease Ontology | 85 |Disease | 86 |10,648 | 87 |- | 88 |- | 89 |https://disease-ontology.org/ | 90 |https://creativecommons.org/publicdomain/zero/1.0/ | 91 |
| DrugBank | 94 |Drug | 95 |15,128 | 96 |Drug-Target, Drug-Enzyme, Drug-Carrier, Drug-Transporter | 97 |28,014 | 98 |https://go.drugbank.com/ | 99 |http://creativecommons.org/licenses/by-nc/4.0/ | 100 |
| Hetionet | 103 |Anatomy, Biological process, Cellular component, Compound, Diease, Gene, Molecular function, Pathway, Pharmacologic class, Side effect, Symptom | 104 |47,031 | 105 |Anatomy–downregulates–Gene, Anatomy–expresses–Gene, Anatomy–upregulates–Gene, Compound–binds–Gene, Compound–causes–Side Effect, Compound–downregulates–Gene, Compound–palliates–Disease, Compound–resembles–Compound, Compound–treats–Disease, Compound–upregulates–Gene, Disease–associates–Gene, Disease–downregulates–Gene, Disease–localizes–Anatomy, Disease–presents–Symptom, Disease–resembles–Disease, Disease–upregulates–Gene, Gene–covaries–Gene, Gene–interacts–Gene, Gene–participates–Biological Process, Gene–participates–Cellular Component, Gene–participates–Molecular Function, Gene–participates–Pathway, Gene→regulates→Gene, Pharmacologic Class–includes–Compound | 106 |2,250,197 | 107 |https://github.com/hetio/hetionet | 108 |https://creativecommons.org/publicdomain/zero/1.0/ | 109 |
| HUGO Gene Nomenclature Committee | 112 |Gene | 113 |41,439 | 114 |- | 115 |- | 116 |https://www.genenames.org/ | 117 |No restriction (https://www.genenames.org/about/) | 118 |
| iDISK | 121 |Dietary Supplement Ingredient, Dietary Supplement Product, Disease, Drug, Anatomy, Symptom, Therapeutic Class | 122 |144,536 | 123 |DSI-Anatomy, DSI-Symptom, DSI-Disease, DSI-Drug, DSI-DSP, DSI-TC | 124 |705,075 | 125 |https://conservancy.umn.edu/handle/11299/204783 | 126 |https://creativecommons.org/licenses/by-sa/3.0/us/ | 127 |
| PharmGKB | 130 |Genes, Variant, Drug/Chemical, Phenotype | 131 |43,112 | 132 |Disease-Gene, Drug/Chemical -Gene, Gene-Gene, Gene-Variant, Disease-Variant, Drug/Chemical-Variant | 133 |61,616 | 134 |https://www.pharmgkb.org/ | 135 |https://creativecommons.org/licenses/by-sa/4.0/ | 136 |
| Reactome | 139 |Genes, Pathway | 140 |13,589 | 141 |Gene-Pathway | 142 |118,480 | 143 |https://reactome.org/ | 144 |https://creativecommons.org/licenses/by/4.0/ | 145 |
| SIDER | 148 |Drug, Side effect | 149 |5,681 | 150 |Drug-Side effect | 151 |163,206 | 152 |http://sideeffects.embl.de/ | 153 |https://creativecommons.org/licenses/by-nc-sa/4.0/ | 154 |
| TISSUE | 157 |Tissue, Gene | 158 |26,260 | 159 |Tissue-Express-Gene | 160 |6,788,697 | 161 |https://tissues.jensenlab.org/ | 162 |https://creativecommons.org/licenses/by/4.0/ | 163 |
| Uberon | 166 |Anatomy | 167 |14,944 | 168 |- | 169 |- | 170 |https://www.ebi.ac.uk/ols/ontologies/uberon | 171 |http://creativecommons.org/licenses/by/3.0/ | 172 |