Re: [bz-biotech] WG: RapidStorm questions from Jun FAN @ Oxford?

[Steve Wolter]

Hi Jun, Markus & Oli in BCC,

thanks for reaching out. The best place to ask rapidSTORM questions is rapidstor...@googlegroups.com. I've added the list to the thread.

The distance threshold is in "mahalanobis distance", i.e., the number of standard deviations of localization uncertainty. Say your localizations has a mean square root deviation of 10 nm, and the existing trace's position has 10 nm uncertainty, then the distance between two of them is expected to be sqrt(2)*10 nm. At distance threshold 5, the algorithm will pair two localizations closer than 5*sqrt(2)*10 nm.

A factor of 5 sounds pretty high. Maybe the localization uncertainty is underestimated?

Best, Steve

On Wed, Jul 3, 2019 at 9:16 AM Reichert Oliver <oliver....@uni-wuerzburg.de> wrote:

Hoffe du bekommst die jetzt nicht von allen hier weitergeleitet ^^

Gruß, Oli

Dear All,

 

kann Jemand die Frage beantworten bitte?

 

Grüße

Cheffe

 

_______________________________________

Prof. Dr. Markus Sauer

Department of Biotechnology and Biophysics

Biocenter

University of Wuerzburg

Am Hubland

97074 Wuerzburg

Germany

Email: m.s...@uni-wuerzburg.de

Phone: +49 931 31 88687

Webpage: www.super-resolution.de

 

Von: Jun Fan [mailto:jun...@physics.ox.ac.uk]
Gesendet: Dienstag, 2. Juli 2019 18:58
An: m.s...@uni-wuerzburg.de
Cc: L-bi...@biozentrum.uni-wuerzburg.de
Betreff: RapidStorm questions from Jun FAN @ Oxford?

 

Dear Prof. Sauer,

 

I'm Jun FAN, a postdoc in Kapanidis group at Oxford. I've been trying to use RapidStorm to get localisation analysis from my fixed-cell data, as there's Kalman Filtering algorithm to remove repeated localisations in consecutive frames. I'm trying to figure out how Kalman Filtering works in the software. I wonder if the distance threshold as illustrated in the attached is related with the filtering? The default value is 2, which doesn't remove repeated localisations. If I change it to 5, then the filtering seems to work. What's the unit for the value please?

I'm trying to contact Dr. Steve Wolter for some information, but just couldn't get the email. 

Do you have some documents or manuals for the RapisStorm that I can check out please? Or you can refer me to your group member who knows the details of software?

Thanks a lot in advance for your kind help.

Kind regards,

 

Jun 

 

Jun FAN, PhD 

Clarendon Laboratory, Condensed Matter Physics

Department of Physics,

Oxford University

Parks Road

Oxford, OX1 3PU

Tel: +44(0)1865 272357

[Steve Wolter]

Dear Jun,

looks correct: Two consecutive localizations closer than 2*sqrt(2)*20nm should be counted as the same molecule, and with mobility set to 0, the molecule's position is the average of all its localizations.

Best, Steve

On Wed, Jul 3, 2019 at 5:33 PM Jun Fan <jun...@physics.ox.ac.uk> wrote:

Dear Steve,

thanks for clarifying the setting for distance threshold. Yes, a factor of 5 is too much, I just use that as an extreme example to check out how Kalman filtering algorithm works.
I have set up sigmax 20nm and sigmay 20nm in track emission. A distance threshold of 2 means take the average of two localizations closer than 2*sqrt(2)*20nm from consecutive frames as one molecule, is that correct?
I guess I have to find a proper value considering localization precision and uncertainty, 2 was just not enough.
Thanks again for your kind reply.
Kind regards,

Jun