Jumping to Conclusions: Reversible jump MCMC developed for change point detection in single molecule localisation microscopy.

Date: Wednesday 12 November 2025, 1.00PM - 2.15PM
Location: Maths & Physics Teaching Centre, QUB (map link provided)
Maths & Physics Teaching Centre, QUB (map link provided)
Local Group Meeting
Book now

Share this event

We have another interesting talk in  the seminar series by Emily Gribbin, Maths & Stats, QUB,  on an medical application involving MCMC Reversible Jump methodology.

This will be a hybrid event with Emily speaking in person in QUB (Maths & Physics Teaching Centre, room G/005, map) and using MS Teams (link here).  
 
 
Fluorescence Localisation Imaging with Photobleaching (FLImP; Iyer et. al., 2024) is used to study clustering of proteins on the cell membrane; a process linked to the development of cancers such as non-small cell lung cancer. FLImP relies on the sequential photobleaching of fluorescent molecules (fluorophores) bound to membrane proteins, producing staircase-like temporal intensity traces as fluorophores switch off one-by-one. However, fluorophores can also display complex photophysical behaviour, including short-lived ‘off’ states (blink-events) and longer-lived dark states, which introduce closely spaced change points that standard methods frequently fail to detect. Currently FLImP uses a heuristic, conservative approach with extensive filtering, often leaving only 1% of all collected data; an inefficiency which makes it unsuitable for clinical use in its current form. By considering the number and locations of these photobleach and ‘off’ state events as a change point problem, we can estimate the number of fluorophores active in each frame; information that, when combined with fluorophore point spread functions, enables spatial localisation and clustering analysis. We present a change point detection algorithm based on reversible jump Markov chain Monte Carlo (Green, 1995), introducing moves to account for short-lived events. Performance is benchmarked against state-of-the-art methods using a comprehensive simulation study and applied to experimental GATTAquant DNA ruler data, demonstrating improved accuracy and increased recovery of usable information for downstream analysis.

Key References
Iyer, R. Sumanth, et al. “Drug-resistant EGFR mutations promote lung cancer by stabilizing interfaces in ligand-free kinase-active EGFR oligomers.” Nature Communications 15 (2024): 2130
Green, Peter J. "Reversible jump Markov chain Monte Carlo computation and Bayesian model determination." Biometrika 82.4 (1995): 711-732.
 
All welcome and we would like to a good turn-out in QUB!

See previous talks here.
Read previous write-ups here.
Follow RSSNI on Linkedin

Gilbert MacKenzie (Chair) &
Hannah Mitchell (Meetings Host)
RSSNI
 
 
Emily Gribbin, Maths. and Stats, Queen's University Belfast.
 
Gilbert MacKenzie 
gilbert.mackenzie.cbs@gmail.com
 
Book now