KdSAXS bayesian

A Bayesian implementation of KdSAXS for calculating dissociation constants from SAXS data for a simple monomer-oligomer model.

Usage

1. Ensure ATSAS software suite is installed and accessible
2. Check the examples below (for Beta-Lactoglobulin protein oligomerization) or upload your own and change the script accordingly:
   • Theoretical data: examples/blg/ph7/theoretical_saxs/
   • Experimental data: examples/blg/ph7/exp_saxs/
3. Update the ATSAS_PATH in the script to point to your ATSAS installation and change other variables
4. Run the script and check example output interactive plots or figures below:

   python kdsaxs_bayesian.py
   

5. Check the KdSAXS repository for more extensive documentation.

Method

The analysis uses Bayesian inference through MCMC sampling to determine protein oligomerization Kd values from SAXS data. The workflow:

1. Initial χ² calculation across a wide Kd range
2. Identification of Kd values where χ² < threshold
3. MCMC sampling within this range using a log-uniform prior
4. Generation of posterior distribution and credible intervals

Optimization

The χ² threshold (default: 2.0) can be adjusted based on the quality of your data. Examine the χ² vs Kd plots to:

• Identify regions of good fit
• Adjust the threshold if needed
• Balance between fit quality and sampling range

Limitations

• Requires manual setting of χ² threshold
• Assumes a simple monomer-oligomer equilibrium
• Limited to single-step oligomerization processes

Dependencies

• Python 3.8+
• ATSAS software suite
• Required Python packages:
  • numpy
  • pandas
  • scipy
  • plotly
  • pymc
  • arviz
  • pytensor

KdSAXS bayesian analysis results example

Interactive Analysis Results download

MCMC example Diagnostics

Interactive Diagnostics download

Analysis Summary

• Kd = 57.30 ± 4.88 µM
• 95% Credible Interval: [49.59, 65.67] µM