Localization Precision Distribution

139 views
Skip to first unread message

Ramunas Stanciauskas

unread,
Jul 14, 2014, 2:47:17 PM7/14/14
to rapidstor...@googlegroups.com
Hello Steve,

First of all thanks for a rather excellent piece of software, I tend to prefer it over other localization software and our lab has been using it consistently for a length of time now. I was curious however, as to obtaining some sort of localization precision distribution. I know in the localization.txt file there is a fit residue value given in chi square, but I'm unsure as to how to translate that into a nanometer precision estimate, or if that's even the correct value to be working with. I know from what I've looked at on github rapidstorm does discard localizations that don't meet a certain fit criteria, so I'm hoping that either I'm missing something in the localization output or another output module that would provide that data. 

And is there anyway to perform a batch type of analysis? I'm working with a large number of region of interests that I'm using localization data to then do several statistical analyses on in matlab, and if I could just queue up rapidstorm to process them in bulk without having to go and try and create a .bat/script it would make things simpler.

Thank you in advance,

Best,

Ram

Steve Wolter

unread,
Jul 15, 2014, 5:16:56 AM7/15/14
to Ramunas Stanciauskas, rapidstor...@googlegroups.com
Hi Ramunas,

if you have supplied camera parameters to rapidSTORM ("Camera response to photon", "Dark intensity"), then the localization precision is computed automatically for 2D fits and stored in the fields sigmaposx/sigmaposy and in the output text file.

In rapidSTORM, you can either queue up jobs in the GUI by changing the file name and clicking "Run" every time, or you can run rapidSTORM from the command line (which is a little involved). Sorry, there's no middle ground, but that's a feature requested somewhat commonly and I'll happily review and accept GitHub pull requests with implementations.

Best regards, Steve


--
You received this message because you are subscribed to the Google Groups "rapidstorm-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rapidstorm-disc...@googlegroups.com.
To post to this group, send email to rapidstor...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Ramunas Stanciauskas

unread,
Jul 15, 2014, 3:54:15 PM7/15/14
to rapidstor...@googlegroups.com, ordin...@gmail.com
Thanks for the response Steve,

I must have missed the camera parameters, easy enough to address. As far as the batch aspect goes I understand, is there any chance you have a brief list of possible command line functions for rapidstorm that I could work off of? In my head I'm thinking to essentially load a reference localization setting file and just write enough command line script to vary the file input/output which shouldn't be too extensive.

Thanks again,

Ram

Steve Wolter

unread,
Jul 15, 2014, 5:29:34 PM7/15/14
to Ramunas Stanciauskas, rapidstor...@googlegroups.com
You can get a (almost full) reference of rapidSTORM options with --help. You can load a settings file with --config. A command line like this should work:

rapidstorm --config foo.txt --inputFile foo.tif --run


Steve Wolter

unread,
Jul 27, 2014, 5:12:57 PM7/27/14
to Ramunas Stanciauskas, rapidstor...@googlegroups.com
2014-07-24 20:42 GMT+02:00 Ramunas Stanciauskas <ordin...@gmail.com>:
That should work from command line as well correct? I tried making a few .bat's but rapidstorm doesn't seem to respond to anything aside from a direct rapidstorm.exe command which opens the program. 

The Windows version of rapidSTORM 3 is compiled in native app mode, which disables any command line interaction. That's the default for wxWidgets apps. rapidSTORM 2 had the command line flags set to allow command-line interaction, but I figured that nobody would ever use the command line on Windows, so I reverted to the default. Seems I was wrong in the assertion, sorry.

The Linux version has command line support as described. 
 
Also, is there a potential way to get the Z localization precision? Or is that confined to just XY currently?

Only XY, and only for 2D fits. For 3D, I've not done the math.
Reply all
Reply to author
Forward
0 new messages