-
Notifications
You must be signed in to change notification settings - Fork 2
/
streamlit_app.py
126 lines (95 loc) · 6.5 KB
/
streamlit_app.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
# DRUG REPURPOSING USING KNOWLEDGE GRAPHS
##########
#LIBRARIES
##########
import streamlit as st
import pandas as pd
#############
#PAGE SET UP
#############
st.set_page_config(page_title="Drug Repurposing",
page_icon=":pill:",
layout="wide",
initial_sidebar_state="expanded"
)
def p_title(title):
st.markdown(f'<h3 style="text-align: left; color:#F63366; font-size:24px;">{title}</h3>', unsafe_allow_html=True)
st.markdown("<h1 style='text-align: center; color: gray; font-size:28px;'>Drug Repurposing using Knowledge Graph Embeddings</h1>", unsafe_allow_html=True)
"""
[](https://github.com/dlopezyse/Drug-Repurposing-using-KGE)
[](https://www.twitter.com/lopezyse)
"""
#########
#SIDEBAR
########
st.sidebar.header('I want to:')
nav = st.sidebar.radio('',['Get drug recommendations :pill:', 'Visualize graph :eyes:', 'About this App'])
st.sidebar.write('')
st.sidebar.write('')
st.sidebar.write('')
st.sidebar.write('')
st.sidebar.write('')
#CONTACT
########
expander = st.sidebar.expander('Contact')
expander.write("I'd love your feedback :smiley: Want to collaborate? Develop a project? Find me on [LinkedIn](https://www.linkedin.com/in/lopezyse/), [Twitter](https://twitter.com/lopezyse) and [Medium](https://lopezyse.medium.com/)")
#######
#PAGES
######
#DRUG RECOMMENDATION
####################
if nav == 'Get drug recommendations :pill:':
st.text('')
p_title('Get drug recommendations')
st.text('')
disease_selection = st.selectbox("Select a disease", ["Dengue", "Chagas", "Malaria", "Yellow Fever", "Leishmaniasis", "Filariasis", "Schistosomiasis"],
help='Select a disease and perform predictions of new drugs on it. This project is focused on seven vector-borne diseases (dengue, chagas, malaria, yellow fever, leishmaniasis, filariasis, and schistosomiasis), but it can be extended to additional ones.')
model_selection = st.selectbox("Select an embedding model", ["TransE", "TransR", "TransH", "UM", "DistMult", "RESCAL", "ERMLP"],
help='Knowledge graph embedding models can encode biological information in a single mathematical space. Select an embedding model and perform drug predictions on the target diseases. This project is focused on seven embedding models (TransE, TransR, TransH, UM, DistMult, RESCAL, and ERMLP), which can be extended to additional ones.')
if st.button("Get recommendations"):
#Drug recommendations
final_selection = disease_selection + model_selection
ranking_file = pd.read_csv('embedding_models/' + final_selection + '.csv', sep = ',')
st.markdown("<h3 style='text-align: left; color:#F63366; font-size:18px;'><b>Drugs recommendations<b></h3>", unsafe_allow_html=True)
st.write(ranking_file)
st.write('The interpretation of the values in the "score" column is model-dependent, and usually it cannot be directly interpreted as a probability.')
st.write('The "in_clinical_trials" column states "yes" if the compound exists in ClinicalTrials.gov for the target diseases, and "no" otherwise.')
#Disease direct connections in DRKG
st.markdown("<h3 style='text-align: left; color:#F63366; font-size:18px;'><b>Direct connections in DRKG for the target disease<b></h3>", unsafe_allow_html=True)
st.write('Find below all entities in GNBR subgraph that are directly linked to the target disease. Compounds are highlighted in red.')
import pandas as pd
import streamlit.components.v1 as components
#Visualize
HtmlFile = open('graphs/knowledge_graph_' + disease_selection + '.html', 'r', encoding='utf-8')
source_code = HtmlFile.read()
components.html(source_code, height = 625,width=750)
#Model performance
performance_file = pd.read_csv('embedding_models/performance_metrics.csv', sep = ';')
# Filter rows where the 'final_selection' column matches final_selection
filtered_data = performance_file[performance_file['final_selection'] == final_selection]
st.markdown("<h3 style='text-align: left; color:#F63366; font-size:18px;'><b>Embedding model performance<b></h3>", unsafe_allow_html=True)
st.dataframe(filtered_data[['Measure', 'Value']], hide_index=True)
#VISUALIZE GRAPH
####################
if nav == 'Visualize graph :eyes:':
st.text('')
p_title('Visualize graph')
st.text('')
number_triples = st.selectbox('Select number of triples to sample from GNBR subgraph', ('50', '100', '250', '500'))
st.write('Zoom in to see graph details')
import pandas as pd
import streamlit.components.v1 as components
#Show GNBR graph
HtmlFile = open('graphs/knowledge_graph_gnbr_filter_' + number_triples + '.html', 'r', encoding='utf-8')
source_code = HtmlFile.read()
components.html(source_code, height = 625,width=750)
#ABOUT THIS APP
###############
if nav == 'About this App':
st.text('')
p_title('About this App')
st.text('')
st.write("Drug repurposing methods can identify already approved drugs to treat them efficiently, reducing development costs and time. At the same time, knowledge graph embedding techniques can encode biological information in a single structure that allows users to operate relationships, extract information, learn connections, and make predictions to discover potential new relationships between existing drugs and vector-borne diseases.")
st.write("In this project, we compare seven knowledge graph embedding models (TransE, TransR, TransH, UM, DistMult, RESCAL, and ERMLP) applied to Drug Repurposing Knowledge Graph (DRKG), analyzing their predictive performance over seven different vector-borne diseases (dengue, chagas, malaria, yellow fever, leishmaniasis, filariasis, and schistosomiasis), measuring their embedding quality and external performance against a ground-truth.")
st.write("This project is based on the paper [Drug Repurposing Using Knowledge Graph Embeddings with a Focus on Vector-Borne Diseases: A Model Comparison](https://link.springer.com/chapter/10.1007/978-3-031-40942-4_8) as developed by [Diego López Yse](https://www.linkedin.com/in/lopezyse/) and [Diego Torres](https://www.linkedin.com/in/ditorres/) for the Conference on Cloud Computing, Big Data & Emerging Topics 2023.")
#-----------------------------------------