Please Note: This repository is the specific pluggin intended to Protac-invent. Source code for our paper "3D Based Generative PROTAC Linker Design with Reinforcement Learning"

Minor modification were conducted for resolving some incompatibility issues of Python environment

In default docking component of Protac-invent, we choose "AutoDock Vina" as docking backends

• 1.The version of AutoDock Vina changed to "AutoDock-Vina-1.2.3" (default: autodock_vina_1_1_2_linux_x86). Because the older version do not support "local only" mode. Two Version also locate in DockStreamCommunity.
• 2.The source code of "DockStream-master/dockstream/core/Autodockvina" is modified.
• 3.The rest of the code can remain unchanged.

DockStream

Description

DockStream is a docking wrapper providing access to a collection of ligand embedders and docking backends. Docking execution and post hoc analysis can be automated via the benchmarking and analysis workflow. The flexilibity to specifiy a large variety of docking configurations allows tailored protocols for diverse end applications. DockStream can also parallelize docking across CPU cores, increasing throughput. DockStream is integrated with the de novo design platform, REINVENT, allowing one to incorporate docking into the generative process, thus providing the agent with 3D structural information.

Supported Backends

Ligand Embedders

• RDKit
• Corina
• OpenEye's OMEGA
• Schrodinger's LigPrep
• TautEnum

Docking Backends

• AutoDock Vina
• rDock
• OpenEye's Hybrid
• Schrodinger's Glide
• CCDC's GOLD

Note: The CCDC package, the OpenEye toolkit and Schrodinger's tools require you to obtain the respective software from those vendors.

Tutorials and Usage

Detailed Jupyter Notebook tutorials for all DockStream functionalities and workflows are provided in DockStreamCommunity. The DockStream repository here contains input JSON templates located in examples. The templates are organized as follows:

• target_preparation: Preparing targets for docking
• ligand_preparation: Generating 3D coordinates for ligands
• docking: Docking ligands
• integration: Combining different ligand embedders and docking backends into a single input JSON to run successively

Using DockStream in REINVENT

DockStream provides a flexible implementation of molecular docking as a scoring function component in REINVENT. The generative agent is able to gradually generate compounds that satisfy the DockStream component, i.e, achieve favourable docking scores. A tutorial notebook is provided.

Requirements

Two Conda environments are provided: DockStream via environment.yml and DockStreamFull via environment_full.yml. DockStream suffices for all use cases except when CCDC GOLD software is used, in which case DockStreamFull is required.

git clone <DockStream repository>
cd <DockStream directory>
conda env create -f environment.yml
conda activate DockStream

Enable use of OpenEye software (from REINVENT README)

You will need to set the environmental variable OE_LICENSE to activate the oechem license. One way to do this and keep it conda environment specific is: On the command-line, first:

cd $CONDA_PREFIX
mkdir -p ./etc/conda/activate.d
mkdir -p ./etc/conda/deactivate.d
touch ./etc/conda/activate.d/env_vars.sh
touch ./etc/conda/deactivate.d/env_vars.sh

Then edit ./etc/conda/activate.d/env_vars.sh as follows:

#!/bin/sh
export OE_LICENSE='/opt/scp/software/oelicense/1.0/oe_license.seq1'

and finally, edit ./etc/conda/deactivate.d/env_vars.sh :

#!/bin/sh
unset OE_LICENSE

Unit Tests

After cloning the DockStream repository, enable licenses, if applicable (OpenEye, CCDC, Schrodinger). Then execute the following:

python unit_tests.py

Contributors

Christian Margreitter (christian.margreitter@astrazeneca.com) Jeff Guo (jeff.guo@astrazeneca.com) Alexey Voronov (alexey.voronov1@astrazeneca.com)