1. Prerequisites and Installation

1.1. Installation

Python: Developed on 3.6 but any version >=3.6 should work

Installing package only:

pip install mdsimseval

Then you can import any method or class

from MDSimsEval.utils import create_analysis_actor_dict  # Example import

Note

Optional: The rmsf_analysis.corr_matrix function in order to provide as ouput a .png requires wkhtmltopdf to be installed via sudo apt-get install wkhtmltopdf. More on the imgkit package. If not installed the output will be an .html file which can be opened with any browser.

Development Environment:

If you are looking for adding functionality or changing the code you will need more packages (eg for the docs).

  1. Clone the repository

  2. Install the requirements.txt

  • Poetry (suggested): poetry install

  • Conda Env: conda install --file requirements.txt

  • VirtualEnv: pip install -r requirements.txt

1.2. Data

As you may have read in the homepage this is not a package for analysis of a single MD simulation. All the methods need a collection of MD simulations preferably of two classes. The goal of the package is to provide insight of which features differentiate these two classes.
Also the simulations must have the same number of frames and preferably performed on the same protein
To make it more clear I will present my use case as an example:
I was provided with a number of 2500 frames simulations on the 5HT2A receptor using different ligands. N of the ligands were agonists and M were antagonists. My goal was to find which features help us differentiate an agonist from an antagonist.
The features I analyzed are:

  • Radius of Gyration

  • Solvent-Accessible Surface Area

  • Root Mean Square Flactuation

  • PCA Loadings and 2D Projections

Continue with reading the data.