InVADo: Interactive Visual Analysis of Molecular Docking Data

paper doi: 10.1109/TVCG.2023.3337642

video doi: 10.1109/TVCG.2023.3337642/mm1

video in Suppl. Mat.: https://ieeexplore.ieee.org/document/10334517/media#media

Replicability Stamp form the Graphics Replicability Stamp Initiative :
http://www.replicabilitystamp.org#https-github-com-marcoschaefert-invado-setup

InVADo_setup - Repository

contains the installation instructions and additional data to complement the main InVADo repository

main repository of InVADo: https://github.com/MarcoSchaeferT/megamol-prolint-InVADo

• InVADo is available under the repository link above
• download the repo and run CMake to prepare InVADo for building
  (see Requirements Building InVADo and Building InVADo)
• running CMake will download the additional content of this repository
• additional content is automatically saved to: InVADo_MAIN_REPO\plugins\prolint\InVADo_setup

additional data includes:

• PLIP installation script
• figures of functional groups (needed for web dashboard)
• small precalculated data set (for web part development or when running the web dashboard without InVADo desktop)
• small molecular docking pipeline
• InVADo config file
• graphics settings file

Requirements Building InVADo:

• operating system: windows 10 or 11

• GPU: NVIDIA

• CUDA (tested CUDA 11.6) https://developer.nvidia.com/cuda-11-6-0-download-archive

• Windows Subsystem for Linux (WSL) with Ubuntu https://learn.microsoft.com/en-us/windows/wsl/install

  • (tested Ubuntu 22.04 LTS) https://www.microsoft.com/store/productId/9PN20MSR04DW?ocid=pdpshare

• CMake (tested 3.26) https://cmake.org/download/

• Visual Studio (tested version of 2022) https://visualstudio.microsoft.com/de/vs/ (VS Version >= 17.10 must be used with CUDA 12.4)

• node.js v16 (tested 16.20.1) https://nodejs.org/en/blog/release/v16.20.0

• Microsoft MPI (install both msmpisdk.msi; msmpisetup.exe) https://www.microsoft.com/en-us/download/details.aspx?id=105289

• Firefox (tested 116.0.1) https://www.mozilla.org/de/firefox/new/

• Python 3.10 (tested 3.10.11) https://www.python.org/downloads/release/python-31011/

  (during Python installation following additional options must be checked [x])

  • Add to Path
  • use --> Customize Installation
  • click 'Next'
  • Precompile Standard Library
  • Download debugging symbols
  • Download debug binaries

• Windows Subsystem for Linux (WSL) https://learn.microsoft.com/en-us/windows/wsl/install

• the rest of the needed programs is downloaded via CMake (except PLIP)

• PLIP runs in the Windows Subsystem for Linux (WSL). PLIP is installed via the script: InVADo_MAIN_REPO\plugins\prolint\InVADo_setup\run_plip_install.cmd (this folder exists after the InVADo_MAIN_REPO is configured by using CMake)

Building InVADo:

• run CMake for MegaMol/InVADo (configure)

  • set visual studio 2022
  • set x64
  

• ignore warnings during the 'configure' process

• check/do the following boxes/steps:

  • ENABLE_MPI
  • set 'MPI_GUESS_LIBRARY_NAME' to 'MSMPI'
  • ENABLE_CUDA
  • BUILD_PROLINT_PLUGIN
  • configure again
  • generate
  • open project
  • (may need to install .NET Framework 4.8.1 SDK)

• set the 'INSTALL' target of 'CMakePredefinedTargets' as the start project

• build and install it as "RELEASE" not "DEBUG"

• "DEBUG" is possible as well, but if the docking data set is not already preprocessed by InVADo as a RELEASE version it will fail to start as a DEBUG version

• the docking data set will be preprocessed one time with the first start of InVADo as RELEASE version

InVADo Configuration

PATH = _InVADo_MAIN_REPO\plugins\prolint\InVADo_setup\

• the InVADo config file InVADoConfig.mmprj is located in the folder 'PATH/InVADo_config'

• in the InVADo config file, the path for the ligand list and the target protein file needs to be updated

• this can be done manually in the XML-based file or graphically as follows:

• manually:

  • open InVADoConfig.mmprj with a text editor and change the path stored in 'value' to e.g.:
  • <param name='pdbqtListFilename' value='C:\PATH\prepare_docking_data_scripts\results\results_list.txt' />
  • <param name="pdbFilename" value="C:\PATH\prepare_docking_data_scripts\7nn9_autoDockTools.pdbqt" />

• graphically:

• the InVADo config can be adjusted with the configurator PATH/InVADo_config/MegaMolConf.exe

• drag InVADoConfig.mmprj and drop it onto MegaMolConf.exe

• an error message will appear -> click ok -> then the following window will be shown:

• click on the first fix button and navigate to _InVADo_MAIN_REPO/build/install/bin/ and select 'mmconsole.exe'

• then click on the fix button of 'Core and Plugin State'

• a new window appears -> there click the button 'Execute'

• after the plugin scan is finished click on 'Use this State file'

• a window appears where a path for saving the state file is already suggested -> use this suggested path and click 'Save'

• close the configurator

• again drag InVADoConfig.mmprj and drop it onto MegaMolConf.exe -> the following view will appear:

• update the paths for the ligand list and the target protein as follows:

  • Module MultiPDBQTLoader1 Parameter: pdbqtListFilename: must be set to your path of a ligand PDBQT file list (*.txt)

    [e.g.:PATH\prepare_docking_data_scripts\results\results_list.txt*]

    ==> please note: 'results' is created after the test docking (see Create a New Data Set)

  • Module PDBLoader1: Parameter: pdbFilename: must be set to your path of a protein file (*.pdb)

    [e.g.: PATH\prepare_docking_data_scripts/7nn9_autoDockTools.pdbqt]

• for easier development set 'INSTALL' target settings of the Visual Studio project as follows:

Create a New Data Set

PATH = _InVADo_MAIN_REPO\plugins\prolint\InVADo_setup\prepare_docking_data_scripts

• the folder 'prepare_docking_data_scripts' contains a pipeline
• this is an exemplary, simple pipeline for processing a molecular docking
• a small example data set is included
  • PATH/data/AAABMN.xaa.pdbqt (includes the ligands)
  • PATH/7nn9_autoDockTools.pdbqt (is the target protein)
• running the script PATH\run_full_automated_docking.cmd automatically performs a molecular docking of the data above
• the docked data is then stored in the folder PATH/results
• the example ligands are from the ZINC database: https://zinc15.docking.org/
• the docking tool is AutoDock Vina: https://vina.scripps.edu/
• if own ligands are used they should follow the ZINC naming scheme

Run InVADo:

PATH = _InVADo_MAIN_REPO\plugins\prolint\InVADo_setup\

• InVADo can be started with mmconsole.exe

• mmconsole.exe is stored after building in "_InVADo_MAIN_REPO/build/install/bin/"

• example command for execution (program + config file): C:\Projects\InVADo\build\install\bin\mmconsole.exe -p "C:\PATH\InVADo_config\InVADoConfig.mmprj" -i Project_1 inst

• the first start can take a while because InVADo preprocesses the data (loading bars in the cmd window will indicate the progress)

• the processed data will be stored in [e.g.: PATH\prepare_docking_data_scripts\results\InVADo_tmpFiles]

Starting only the Web Dashboard

PATH = _InVADo_MAIN_REPO\plugins\prolint\server\

• there is the possibility to run only the web part of InVADo without the 3D visualization
• it will use the provided test data set
• build the web app: run 'PATH\build_app.cmd'
• start the web app: run 'PATH\run_app.cmd'

optional: improve rendering quality

PATH = _InVADo_MAIN_REPO\plugins\prolint\InVADo_setup\

• got to folder 'graphicSettings' located in PATH
• run setGraphics.cmd to set an Nvidia profile for InVADo

Trouble Shooting

• if the build fails with: can not find _Py_wfopen() make sure Python 3.10 is installed and no other Python version is set for the Windows PATH variable