#!/usr/bin/env python # coding: utf-8 # In[ ]: import streamlit as st import pandas as pd from PIL import Image import subprocess import os import base64 import pickle # Logo image image = Image.open('logo.png') st.image(image, use_column_width=True) st.markdown(""" # Vitamin D Receptor (VDR)-Pred VDR-Pred is a biological activity prediction tool using a machine learning-based QSAR model. **Credits** - App built in `Python` + `Streamlit`, by Centre for Research in Molecular Modeling (CERMM), Concordia University, Montreal, Quebec, Canada, H4B 1R6 - Descriptor calculated using [PaDEL-Descriptor](http://www.yapcwsoft.com/dd/padeldescriptor/) [[Read the Paper]](https://doi.org/10.1002/jcc.21707). """) # In[ ]: # Navigation bar rad=st.sidebar.selectbox("Menu", ["Home", "Help", "Team", "Terms of Services"]) if rad =="Home": st.title ('Home') st.header ('About VDR-Pred') st.markdown(""" Vitamin D Receptor (VDR) is a drug target for many infectious and immunological diseases. VDR-Pred tool employs to predict VDR agonistic activity (i.e. pIC50) of drug-like molecules using Machine learning algorithm-based QSAR models. The screening results allow the user to use the predicted molecules to activate the vitamin D receptor enzyme. """) st.header ('Inhibitory Concentration-50 (IC50)') st.markdown(""" The half-maximal inhibitory concentration (i.e. IC50) is defined as a quantitative measurement of how much amount of an actual inhibitory molecule (i.e. drug or chemical) is required to inhibit 50 % of targeted biological activities (i.e. cell growth, enzymatic action, and biological pathways). """) st.header ('pIC50') st.markdown(""" pIC50 is the negative (-) logarithmic value of IC50 while converting into molar (Ex. IC50 of 1µM is 10-6M, the pIC50 is 6.0). The advantage of pIC50 is to think about the potency of drug-like compounds, and it is essential for dealing with concentration-dependent relationships. """) if rad =="Help": st.title ('Help') st.header ('Steps to calculate the pIC50') col1, mid, col2 = st.columns([1,1,200]) with col1: st.header ('Steps 1-2') with col2: st.image('Step 1.jpg', width= 500) st.markdown(""" **Step 1:** Upload a .txt file with “SMILES” of small molecules that are selected from CHEMBL, PubChem, or DrugBank or ZINC database. The uploaded .txt file must contain a “SMILES” and “ID” of selected small molecules. The SMILES and reference ID (Ex:`C=C1C[C@H](C[C@@H](C)' and ‘CHEMBL100’) must be written IUPAC standard format. User may avoid non-standard format and characters. **Step 2:** Click Submit button. The calculation may take few seconds to 15 minutes based on the number of compounds. """) col1, mid, col2 = st.columns([1,1,200]) with col1: st.header ('Steps 3-4') with col2: st.image('Step 2.jpg', width= 500) st.markdown(""" **Step 3:** User can visualize the list of uploaded input molecules detail (i.e. SMILE and ID) in tabular form. **Step 4:** User can visualize the calculated molecular descriptors of the input molecules in tabular form. """) col1, mid, col2 = st.columns([1,1,200]) with col1: st.header ('Steps 5-7') with col2: st.image('Step 3.jpg', width= 500) st.markdown(""" """) st.markdown(""" **Step 5:** User can visualize Subset of molecular descriptors from previously build QSAR model in tabular form . **Step 6:** The calculated molecular descriptors and pIC50 values displayed in tabular form. **Step 7:** User can download the results in .csv format. """) if rad == "Team": st.title ('Team') st.markdown( "**Prof. Gilles H. Peslherbe, PhD**, \n" "Principal Investigator & Director of CERMM \n" "CERMM, Concordia University, Montreal, Canada \n" "**Dr. Gurudeeban Selvaraj, PhD**, \n" "Research Associate & Developer \n" "CERMM, Concordia University, Montreal, Canada \n" "**Dr. Satyavani Kaliamurthi, PhD**, \n" "Horizon Postdoctoral Fellow & Co-Developer \n" "CERMM, Concordia University, Montreal, Canada \n" "**Dr. Denise Koch, PhD**, \n" "CERMM Manager & Co-Developer, \n" "CERMM, Concordia University, Montreal, Canada \n" "**Prof. Dongqing Wei, PhD**, \n" "Professor & International Collaborator, \n" "College of Life Sciences and Biotechnology, \n" "Shanghai Jiao Tong University, Shanghai 200240, China " ) st.title ('Contact') st.markdown( "CERMM, \n" "Concordia University, 7141 Sherbrooke West Street., \n" "Montreal, Quebec, Canada H4B1R6 \n" "Phone: XXXXXXXX \n" "Email: XXXXXXXX" ) st.title ('Acknowledgements') st.markdown( "Natural Sciences and Engineering Research Council (NSERC) Discovery Grant, Canada \n" "Horizon Postdoctoral Fellowship, Concordia University, Montreal, Quebec, Canada" ) if rad == "Terms of Services": st.title ('Terms of Services') st.markdown('* This VDR-Pred is available to you subject to the Terms of Service described on this page. Your use of this site constitutes your agreement to these terms and conditions. \n' '* The contents of the VDR-Pred are intended for educational or research purposes. \n' '* All materials and services provided on the Web application are protected by copyright or other intellectual property rights, and owned or controlled by the CERMM, Concordia University.') # Molecular descriptor calculator def desc_calc(): # Performs the descriptor calculation bashCommand = "java -Xms2G -Xmx2G -Djava.awt.headless=true -jar ./PaDEL-Descriptor/PaDEL-Descriptor.jar -removesalt -standardizenitro -fingerprints -descriptortypes ./PaDEL-Descriptor/PubchemFingerprinter.xml -dir ./ -file descriptors_output.csv" process = subprocess.Popen(bashCommand.split(), stdout=subprocess.PIPE) output, error = process.communicate() os.remove('molecule.smi') # File download def filedownload(df): csv = df.to_csv(index=False) b64 = base64.b64encode(csv.encode()).decode() # strings <-> bytes conversions href = f'Download Predictions' return href # Model building def build_model(input_data): # Reads in saved regression model load_model = pickle.load(open('vdr_model.pkl', 'rb')) # Apply model to make predictions prediction = load_model.predict(input_data) st.header('**Prediction output**') prediction_output = pd.Series(prediction, name='pIC50') molecule_name = pd.Series(load_data[1], name='molecule_name') df = pd.concat([molecule_name, prediction_output], axis=1) st.write(df) st.markdown(filedownload(df), unsafe_allow_html=True) # Sidebar with st.sidebar.header('Upload your data'): uploaded_file = st.sidebar.file_uploader("Upload your input file", type=['txt']) st.sidebar.markdown(""" [Example input file](https://github.com/Gurudeeban-Selvaraj/advanced-therapeutics/blob/main/example.txt) """) if st.sidebar.button('Submit'): load_data = pd.read_table(uploaded_file, sep=' ', header=None) load_data.to_csv('molecule.smi', sep = '\t', header = False, index = False) st.header('**Original input data**') st.write(load_data) with st.spinner("Calculating descriptors..."): desc_calc() # Read in calculated descriptors and display the dataframe st.header('**Calculated molecular descriptors**') desc = pd.read_csv('descriptors_output.csv') st.write(desc) st.write(desc.shape) # Read descriptor list used in previously built model st.header('**Subset of descriptors from previously built models**') Xlist = list(pd.read_csv('descriptor_list.csv').columns) desc_subset = desc[Xlist] st.write(desc_subset) st.write(desc_subset.shape) # Apply trained model to make prediction on query compounds build_model(desc_subset) else: st.info('Contact administrator!')