Skip to content

Protein Docking

This application is modified from ParslDock.

Protein docking aims to predict the orientation and position of two molecules, where one molecule is a protein, and the other is a protein or a smaller molecule (ligand). Docking is commonly used in structure-based drug design, as it can predict the strength of docking between target small molecule ligands (drugs) to target binding site (receptors).

Installation

The docking application requires Conda to install all of the required dependencies.

conda create --name taps-docking python=3.11
conda activate taps-docking
conda install -c conda-forge -c bioconda autodock-vina libglu mgltools vmd
python3 -m pip install -e .[docking]

Warning

This environment installs Python 2 and 3 so python3 will need be be used for all commands.

The MGLTOOLS_HOME environment must be set. If you install MGLTools with Conda, the path is likely the same as CONDA_PREFIX. 

export MGLTOOLS_HOME=$CONDA_PREFIX

Warning

Dependencies are not compatible with ARM64 architectures.

Tip

Certain executors do not play nicely with the parallelism used by the simulation codes. If tasks appear to get suck, it may be necessary to set OMP_NUM_THREADS=1.

Data

Sample input data is provided in the ParslDock tutorial. The following CLI will download the dataset_orz_original_1k.csv (SMILES string), 1iep_receptor.pdbqt (input receptor), and set_element.tcl (TCL script) files.

python3 -m taps.apps.docking.data --output data/docking/

Example

The docking application can be invoked as follows.

python3 -m taps.run --app docking \
    --app.smi-file-name-ligand data/docking/dataset_orz_original_1k.csv \
    --app.receptor data/docking/1iep_receptor.pdbqt \
    --app.tcl-path data/docking/set_element.tcl
    --engine.executor process-pool \
    --engine.executor.max-processes 40 \

Checkout the full list of docking parameters with python -m taps.run --app docking --help. For example, the --app.batch-size and --app.num-iterations parameters control the parallelism and length of the application.

Failure

If you get an error that looks like the following: 

ImportError: /home/cc/miniconda3/envs/taps-docking/lib/python3.11/lib-dynload/_sqlite3.cpython-311-x86_64-linux-gnu.so: undefined symbol: sqlite3_trace_v2
This is because Python is linking against an incompatible version of libsqlite. You may need to install libsqlite yourself and/or set LD_LIBRARY_PATH to prefer the right version. If you are following the installation instructions here, we found the following to work. 
conda install "libsqlite<3.46.0"
See Issue #151 for further discussion and debugging tips.