basic questions

[Bernd Zobiak]

Hi,

I am new to the rapidSTORM software and have some basic questions:

1) File length: can I overcome the limit of 4GB TIFF files? rapidSTORM cannot read bigTIFF files

2) When I calculate the no. of images in the file, the output is not correct (e.g. calculated 6765 images in a file of 10000). Will RS use all the images or only the calculated no. for reconstruction then?

3) ADC: Do I understand correctly, this is simply the gray value?

4) Drift correction: the term "p(fr^-1 m)" does it mean picometer per frame?

Sorry for bothering you with these basic questions but I did not find a helpful answer in the manual.

Thanks for any help and best wishes,

Bernd

[Steve Wolter]

Hi Bernd,

I'm currently very busy and my answer might take a few weeks. Sorry. I've scheduled it for my next train ride (2015-07-08). List, can anyone help earlier?

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.

[Sven Proppert]

Hi Bernd,

I hope the following will help you:
1) Usually, the file system (Windows) is limited to 4GB files. Thus, most camera softwares split the movies after 4GB but you can join files in rapidSTORM. In order to do so, you have to set the "number of input layers" at the very top of the rapidSTORM window to the appropriate number. Let's assume you work with the Andor Solis software you will have filesnames "foo.tif", "foo_X2.tif" and "foo_X3.tif". Set the "number of input layers" to 3. Now, set "join inputs of" to "in time" (this is the first control below the input box. All you have to do now is to load the split files in order ("foo.tif" in layer1, "foo_X2.tif" in layer 2 and so on). rapidSTORM will now process the entire measurement accordingly.
2) This is a known bug and was reported before - I have to admit that this bug report is a bit hard to find in this newsgroup... Anyway, if you have a look at your localizations file, usually the third column is the frame number and when you scorll down till the end of the file you should see, that the number is correct, even though "determine length of file" stopped somewhere at a wrong number. Do not be confused: rapidSTORm counts the frames starting with frame 1 being indexed with 0. So the last number of the third column should be 9999 for a 10.000 frames movie, if there was at least one localization in the last frame.
3) In my understanding, you are right. ADC means "analog to digital counts" and I would say, this equals the gray value.
4) Yes, it means picometer per frame. Please be aware that the linear drift correction is corrupted - I will shortly report the bug separately... To cut it short, the linear drift correction works fine but gives wrong values. If  for example you need to subtract a linear drift of 1 pm/fr to get a drift-free image, the real value subtracted is 3. In other words: If you did the drift correction manually with a data handling software like Origin, you would need to subtract a drift of 3 pm/fr in order to get the same result.
This should not bother you too much in most cases but becomes a problem when doing sequential two color dSTORM because in order to get colocalization of corresponding signals, you need to subtract the drift that occured in the first measurement thom the found localizations of the second color. If you would subtract the value estimated with the linear drift correction module (1 pm/fr * 10.000 frames), your data will be shifted because you should have subtracted 3 * 1 pm/fr * frame. When entering the correct value, the measurements will give the correct result.

Please don't hesitate to ask in case of further uncertainties.

Best,
Sven

[Bernd Zobiak]

Hi,

thanks a lot for you explanations! 

Regarding the file limitation: I am able to generate files larger than 4GB (default file format is .vsi from Olympus CellSens, but I can also create also *.btif). So even if single files are larger than 4GB, the only way in RS is to use split files that are smaller and merge them? Or can RS read SIF files that are larger? Then I might convert them somehow (in ImageJ maybe). Otherwise I will make us of the merge function.

Q: Is RS able to make a non-linear drift correction? Like a function that allows a bead track or so to use for correction would come to my mind (this bUnwarpJ functionality I unfortunately don't know at all how to use it).

Thanks and cheers

Bernd

[Sven Proppert]

Hi,

I honestly do not know, whether RS is capable of dealing with larger files as I only used Andor Solis or Micromanager for image acquisition and they both split automaticall after 4 GB... But I think, as RS is a 64bit software, it should be ok...? If I was you, I would just try to load a larger file to RS and see what happens. As I mentioned before, assuming RS runs and doesn't crash due to the large files, you can check in the localizations file ("name of your input/measurement".txt) by scrolling to the very bottom and checking the value in the third column (as long as you measure in 2D). That number is the index of the last frame processed (-1). If it fits to the length of your measurement, everything is fine.
I would rather not convert large movie files because it takes A LOT of (computational) time...

Q: Is RS able to make a non-linear drift correction?

A: If you have fiducials in your sample, you can use a two-step drift correcftion approach:
1. Extract traces of the fiducials from your measurement.
2. Apply that bead trace(s) as nonlinear drift correction.
Unfortunately, I did not yet do it myself. You could have a look here: https://groups.google.com/forum/#!topic/rapidstorm-discuss/D1YHJu6cOrc
and here: https://github.com/stevewolter/rapidSTORM/commit/00b269aee191c062d2b42827f7f2e3bcbc9d3d87
I hope, this gives you some help. If not, please supply us with some small exemplary data, that we can fiddle around with.

Cheers,
Sven

PS: bUnwarpJ is not the plugin you are looking for. It does: unwarp/unstretch an image in order to fit to another image. (Their manual website helps quickly getting an idea...) It does not: Fiducial tracking and drift subtraction...

[Bernd Zobiak]

Hi,

I tested larger files and RS will give an error for those files, but no problem, I can deal with that and the workaround you suggested.

Regarding the drift correction: I found a combined approach posted by Yoshiyuki Arai on Github (https://github.com/arayoshipta/rapidSTORMdriftCorrection). It uses two RS localization files, 1 from a fiducial and 1 from the target image (both images will be processed in RS), and applies a macro to create a drift corrected image in ImageJ. This worked very well in my hands today. I keep testing it with more datasets but I think this is a good way to proceed.

Cheers,

Bernd