Application of PSO to enzymatic mechanism

Code to support paper in https://www.cell.com/patterns/

Application of PSO (Particle Swarm Optimization) to understand the allosteric mechanism of action of inhibition of the enzyme HSD17ß13

Requirements:

• Python 3.9

Instructions

To perform a single run, install the requirements and run the main run.py python file.

pip install -r requirements.txt
python run.py

Overview of code

• Fluorescence data for a “zero trace” (i.e., protein alone, no inhibitor) is analysed and parameters are optimized using PSO.
• Linear gradient descent is performed on the “zero trace” following PSO to refine fitting parameters
• PSO is used on the individual traces (i.e., each inhibitor concentration) of thermal shift data to derive local parameters
• Then, optimal global parameters for all traces are determined using PSO and refined using linear gradient descent

Expected results

You will see the pyswarms output describing the progress of each PSO step. At the end, 2 plots should be produced of the final 2 fitting steps, the latter looking like the plot below:

Note: A perfect fit is not expected to occur every time. For this dataset, it should reach the correct final solution 15% of the time. See below.

Configuration:

• Multiple runs: If you would like to see the performance of the model over a set of runs, increase NUMBER_OF_RUNS in run.py to e.g. 100.
• Saving images/files: If you would like the results of the fits exported to csvs, and the files and plots generated saved to disk, create the export folders CSV Exports and Plots in the home directory and set SAVE_TO_DISK in run.py to True