Bayesian detection of intensity changes in single molecule and molecular dynamics trajectories

Area of Science:

• Physical Chemistry
• Computational Chemistry
• Biophysics

Background:

• Single molecule spectroscopy and molecular dynamics simulations track molecular dynamics over time.
• Analysis involves investigating changes in followed degrees of freedom, such as photon counts in fluorescence experiments.

Purpose of the Study:

• Introduce a straightforward Bayesian method for change point detection in single molecule data.
• Provide a systematic approach for identifying when and if changes in molecular dynamics occur.

Main Methods:

• Developed a Bayesian method for analyzing time-series data from single molecule experiments.
• Applied the method to detect changes and cluster data into distinct states.
• Focused on Poisson-distributed data but noted applicability to other distributions.

Main Results:

• The Bayesian method offers a systematic approach to change point detection.
• Enables effective clustering of data into different molecular states.
• Provides a simpler framework for hypothesis testing compared to maximum likelihood methods.

Conclusions:

• The proposed Bayesian method is effective for analyzing single molecule dynamics.
• It offers advantages in change point detection, state clustering, and hypothesis testing.
• The methodology has broad applicability to various data distributions in physical sciences.