pydock3

pydock3 is a Python package wrapping the Fortran program UCSF DOCK that provides tools to help standardize and automate the computational methods employed in molecular docking. It is a natural successor to DOCK Blaster, originally published in 2009, and blastermaster, part of the DOCK 3.7 release in 2012.

Documentation: https://wiki.docking.org/index.php/Pydock3 DOCKopt documentation: https://wiki.docking.org/index.php?title=Dockopt_(pydock3_script)

Installation

# Through poetry
conda create -n pydock3 python=3.9
conda activate pydock3
pip install poetry
bash install.sh

How to run dockopt

0. prepare

A job_dir_path is needed where dockopt runs. actives.tgz and decoys.tgz are zip file of directory of db2 files. The structure of job_dir_path and tgz files are as below: NOTE: dockopt_config.yaml is control file of dockopt. If not given, will copy pydock3/dockopt/default_dockopt_config.yaml instead. The yaml file should have the same format as in pydock3/dockopt/dockopt_config_schema.yaml

job_dir_path/
    - (optional) dockopt_config.yaml
    - rec.pdb
    - xtal-lig.pdb
    - actives.tgz
    - decoys.tgz

actives.tgz -> actives/
decoys.tgz -> decoys/
actives/
    - ligands/
        - 1.db2
        - 2.db2
        ...
decoys/
    - ligands/
        - a.db2
        - b.db2
        ...

1. run

through python script (Recommended)

python run_dockopt.py \
    --schedular "slurm" \
    --job_dir_path "/your/path/" \
    --tmp_dir "/tmp/your/tmp/path" \
    --config_file_path "dockopt_config.yaml" \
    --actives_tgz_file_path "actives.tgz" \
    --decoys_tgz_file_path "decoys.tgz" \
    --recpdb "rec.pdb" \
    --ligpdb "xtal-lig.pdb" \
    --extra_submission_cmd_params_str "-p honda -N 1 -n 1 -q normal" \
    --retrodock_job_timeout_minutes "12:00:00" \
    --retrodock_job_max_reattempts 0 \
    --max_task_array_size 20000 \
    --dock38bin "$DOCKBASE/docking/DOCK/bin/dock64"

Or, through command line

export SBATCH_EXEC="/usr/bin/sbatch"
export SQUEUE_EXEC="/usr/bin/squeue"
export DOCK3_EXECUTABLE_PATH="$DOCKBASE/docking/DOCK/bin/dock64"
export TMPDIR="/tmp/qcxia"

pydock3 dockopt - run slurm \
--job_dir_path="/your/path/" \
--config_file_path="dockopt_config.yaml" \
--actives_tgz_file_path="actives.tgz" \
--decoys_tgz_file_path="decoys.tgz" \
--retrodock_job_max_reattempts=0 \
--allow_failed_retrodock_jobs=False \
--retrodock_job_timeout_minutes="12:00:00" \
--max_task_array_size=None \
--extra_submission_cmd_params_str="-p honda -N 1 -n 1 -q normal" \
--sleep_seconds_after_copying_output=0 \
--export_decoys_mol2=False \
--delete_intermediate_files=False \
--force_redock=False \
--force_rewrite_results=False \
--force_rewrite_report=False

Multiple dockfiles will be generated locally and multiple docking campaigns will be submited remotely (through slurm as array job). Running status can be monitored from log (for python running, an example.log will be recorded as screen outputs).

2. analysis

After running, results will be given as a directory best_retrodock_jobs and 2 files report.html, result.csv. best_retrodock_jobs are top scored docking campaigns. dockfiles within can be used afterwards. result.csv is detailed record of dockfiles and results. report.html gives meaningful plot of dockopt results.

---

TIPS:

1. Dockfiles generation (blastermaster) is run on local machine (without parallel), which may be time-consuming if the combination number is huge. Should think about move this part to remote as well.