DEEPScreen: Virtual Screening with Deep Convolutional Neural Networks Using Compound Images

• Important notice: This is the new version of DEEPScreen developed using PyTorch framework, please go the master branch of this repository to reach the old version and other information presented in the paper. We advice to use this new version of DEEPScreen to train target-specific models. Please note that this version is planned to be improved further by adding more functionalities.
• DEEPScreen is a large-scale DTI prediction system, for early stage drug discovery, using deep convolutional neural networks
• One of the main advantages of DEEPScreen is employing readily available 2-D structural representations of compounds at the input level instead of conventional descriptors that display limited performance
• DEEPScreen learns complex features inherently from the 2-D representations, thus producing highly accurate predictions.
• More information can be obtained from DEEPScreen journal article.

Installation

DEEPScreen is a command-line prediction tool written in Python 3.7.1. DEEPScreen was developed and tested in MacOSx but it should run in any Unix-like operating system. Please run the below commands to install requirements for model training and testing. Dependencies are available in requirements.txt file which is located under bin directory.

conda create -n deepscreen_env python=3.7
source activate deepscreen_env
pip install -r requirements.txt

Descriptions of folders and files in the DEEPScreen repository

• bin folder includes the source code of DEEPScreen.

• training_files folder includes the files directly used in the training and testing of the system:

  • chembl27_preprocessed_filtered_bioactivity_dataset.tsv.zip updated version of ChEMBL preprocessed and filtered dataset contains drug/compound-target interactions from the ChEMBL database (v27) after the application of multiple filtering operations to obtain a clean training set,

  • chembl27_training_target_list.txt list of target chembl ids,

  • target_training_datasets contains a folder (e.g. CHEMBL286) for each target where each target folder contains

    • a json file named train_val_test_dict.json which includes train/validation/test compound ids,
    • a folder named imgs which holds images of compounds.

  • chembl27_preprocessed_filtered_act_inact_comps_10.0_20.0_blast_comp_0.2.txt contains the active and inactive compound information for each target protein in ChEMBL, after the similarity-based negative training dataset enrichment process. In this file, there are two lines for each target, in the following format:

    CHEMBL286_act    CHEMBL1818056,CHEMBL2115367,CHEMBL344651,CHEMBL62054, ...
    CHEMBL286_inact  CHEMBL288434,CHEMBL584926,CHEMBL406111,CHEMBL151055, ...

    The list of active/inactive compounds separated by commas (i.e., the second tab seperated column: CHEMBL1818056,C...) for the correnponding target (i.e., the first column: CHEMBL286_act),

  • chembl27_uniprot_mapping.txt contains the id mapping between UniProt accessions and ChEMBL ids for proteins, in tab-separated format (Target UniProt accession, Target ChEMBL id, Target protein name and Target type),

• result_files folder contains results of various tests/analyses:

• 2-D images of:

  • 409,311 ChEMBL compounds in the train/validation/test datasets of 812 target proteins of DEEPScreen can be downloaded from here
  • all compounds (~2M) in ChEMBL v27 can be downloaded from here
  • all drugs (~11K) in DrugBank v5.1.7 can be downloaded from here

How to train DEEPScreen models and get performance results

• Clone the Git Repository

• Download the compressed file for the target that you want to train here

• Locate the zipped target file under training_files/target_training_datasets and unzip it

• Run the main_training.py script as shown below

Explanation of Parameters

• --targetid: Target to be trained (default: CHEMBL286)

• --model: CNN architecture to be used (default: CNNModel1)

• --fc1: number of neurons in the first fully-connected layer (default:512)

• --fc2: number of neurons in the second fully-connected layer (default:256)

• --lr:learning rate (default: 0.001)

• --bs: batch size (default: 32)

• --dropout: dropout rate (default: 0.1)

• --epoch: number of epochs (default: 200)

• --en: the name of the experiment (default: my_experiment)

To perform training for a target (CHEMBL286 in the below example):

python main_training.py --targetid CHEMBL286 --model CNNModel1 --fc1 256 --fc2 128 --lr 0.01 --bs 64 --dropout 0.25 --epoch 100 --en my_chembl286_training

Output of the scripts

main_training.py creates a folder named (given as argument --en) under result_files/experiments folder. Two files are created under results_files/experiments/:

• best_val_test_predictions--.txt contains predictions for independent test dataset.
• best_val_test_performance_results--.txt which contains the best test performance results. Sample output files for ChEMBL286 target is given under results_files/experiments/my_chembl286_training.

Article

If you use DEEPScreen please consider citing:

Rifaioglu, A. S., Nalbat, E., Atalay, V., Martin, M. J., Cetin-Atalay, R., & Doğan, T. (2020). DEEPScreen: high performance drug–target interaction prediction with convolutional neural networks using 2-D structural compound representations. Chemical Science, 11(9), 2531-2557.

License

DEEPScreen Copyright (C) 2020 CanSyL

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.