Skip to content

Installation

Installing MIPkit

Installation of MIPkit is easy, but requires a few things to be installed first.

It is also highly recommended that CUDA or ROCM is also installed, and GROMACS compiled support for GPU acceleration.

To install MIPkit, first create a virtual environment with python -m venv MIPvenv. For future use, create a virtual environment activation bash script with :

source ~/path/to/MIPvenv/bin/activate
# Through amber
source ~/path/to/AmberTools24/amber24/amber.sh
# Or
source ~/path/to/pmemd24/amber.sh
source ~/path/to/ambertools25/amber.sh

This will do two things: First, it will start the virtual environment. Second, it will add the ambertools libraries to the PATH so that we can use them with Acpype.

Finally,

git clone https://github.com/tj-barrett/BMBT-MIPkit/
cd BMBT-MIPkit
pip install BMBT-MIPkit/.

Once installed, MIPkit -init to generate sdfs for docking. This will only need to be run again if the FM list is updated.

Note : If Vina, GROMACS, and AMBER are not in their default locations, it might be necessary to change the MIPkit/config/config.yaml locations to reflect the correct bin locations.

Potential Installation Errors

  • acpype : You might need to install libhdf4-dev, libhdf5-dev first. On Ubuntu 20.04 we experienced an issue with this. Using apt install for both dev packages, then recompiling and installing the Ambertools binary solved the problem.
  • acpype : If acpype fails, there might be an issue with the atom typing. Constants.py contains a list of specific-generic conversions, and that list might need to be updated if there is an atom type specified in the pdb that is not recognized by acpype.
  • GNINA : Installation on Ubuntu is much more straightforward than on RHEL (Centos)
  • GROMACS : MIPkit was built using 2024.6, so it is recommended. Newer versions should work as well.