sample data?

[Jessie Weber]

Hello,

First of all, thank you for making quickpalm available! I'm having a
lot of fun with it.

I was wondering if you would be willing to provide any sample data for
fitting. I would like to test if I can generate the same results you
get for any given sample data set.

Best,
Jessie

[Ricardo Henriques]

Hi Jessie,

Sure, what kind of data would you like, 2D or 3D? Note that it will
have to be a very small dataset in order to fit in an email or online
upload. I will only be able to do this next week - I'm away from my
data workstation meanwhile.

On the mailing list there is also a dataset of beads under astigmatism
that you can also use -
http://groups.google.com/group/quickpalm/files?hl=en_US.

Cheers,
R

--
Ricardo Henriques
PhD Student
Gene Expression and Biophysics Unit
Institute of Molecular Medicine
Faculty of Medicine, University of Lisbon
Av. Prof. Egas Moniz
1649-028 Lisbon, Portugal
Phone: + 351 217999503, Ext: 47318
Fax: + 351 217999504
E-mail: rhenr...@fm.ul.pt

[Jessie Weber]

Hi Ricardo,

Thank you for your fast response! The beads already in the mailing
list are helpful, but I'm looking for a mulit-spot 2D image to play
with. I previously generated a line of 10 Gaussian points for testing
but now would like to move to something more complicated with
signals/distributions found in real data.

Thank you again for your help!

Best,
Jessie

--

Jessie Weber, PhD
Postdoctoral Fellow
Montreal Neurological Institute/Department of Biology
McGill University
jessie...@mcgill.ca
jessie....@gmail.com

[Ricardo Henriques]

Hi Jessie,

I have uploaded an example 4MB 555x64x64 example dataset that you can
use - https://sites.google.com/site/quickpalmtutorials/SampleSmallNeuronDendrite.tif.
It is a small portion of a dendrite labeled with Cy3 taken under
dSTORM. You will notice that the first 2 frames are brighter - this
corresponds to the initial bleach step.

For gaussian fitting I have had great results with the MTT algorithm
also (matlab, open-source)-
http://www.nature.com/nmeth/journal/v5/n8/full/nmeth.1233.html.

Hope this helps,
R

[Ann Wheeler]

Hi Ricardo

Thanks for uploading the test dataset and the nature methods paper.
We've used it too! Very handy.

Best wishes

Ann

Dr Ann Wheeler
Blizard Advanced Light Microscopy Facility Manager
Blizard Institute of Cell and Molecular Science
Barts and The London School of Medicine and Dentistry
Queen Mary University of London
Blizard Building
4 Newark Street
London E1 2AT

Phone: 020 7882 2406
Fax: 020 7882 2180

[Jessie Weber]

Hi Ricardo,

Thanks again for your fast response! I think this data set should be
perfect for testing. However, when I click the link, a single tif
comes up in a new webpage. Perhaps I'm missing something obvious, but
could you clarify how to download the image stack?

Thank you very much for your help!

Best,
Jessie

[Ricardo Henriques]

Hi Jessie,
Right click on the link and download the file, then open in ImageJ and
you will see that it is a multi-frame tiff file.
Let me know if you have any problems.
Cheers,
R

[Sebastian Rhode]

Hi Ricardo,

I am also not able to download the multi-page TIFF. If I open the file using ImageJ there is only 1 page (311Bytes file size) ...

Cheers,

Sebi

2011/2/7 Ricardo Henriques <paxc...@gmail.com>

--
Dr. Sebastian Rhode
Grünwalder Str. 103a
81547 München
Tel: +49 89 4703091
Mobil: +49 15122810945
sebr...@googlemail.com

[Jessie Weber]

Hi Ricardo,

It's still just appearing as a 64x64 8-bit single black image. The
downloaded tif is only 311 bytes.

Thanks,
Jessie

[Ricardo Henriques]

Hello Jesse and Sebastian,

Sorry for this, I've uploaded the file again to the QuickPALM site, you should be able to see it in the downloads area - http://code.google.com/p/quickpalm/downloads/list . Let me know if it works.

Cheers,

R

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Ricardo Henriques

Gene Expression and Biophysics Unit  
Institute of Molecular Medicine
Faculty of Medicine, University of Lisbon Av. Prof. Egas Moniz 
1649-028 Lisbon, Portugal 
Phone: + 351 217999503, 
Ext: 47318 
Fax: + 351 217999504 
E-mail: rhenr...@fm.ul.pt

~~ Or ~~ 
Unité de Biochimie Structurale et Cellulaire                          
Département de Biologie Structurale et Chimie
Institut Pasteur                          
25 rue du Docteur Roux                          
75724 Paris Cedex 15       
Phone: +33 1 40 61 31 70
Ext: 3170
E-mail: rica...@pasteur.fr

[Jessie Weber]

Thanks, Ricardo, that dowload works perfectly and is the full stack!

Best,
Jessie

[Jessie Weber]

Hi Ricardo,

Thank you for providing the sample data back in February. I'm just
looking at it again now and I'm wondering how I can know if I'm
generating appropriate images. Do you have recommended settings for
this dataset? Is the final image from this data published anywhere for
comparison?

Again, thank you so much for sharing your work so openly with the
imaging community!

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Ricardo Henriques]

Hi Jessie,

No problem. Here's my settings advice:
- the image pixel size is 106.667nm (16um camera, 100x objective, 1.5x
tube lens)
- for this dataset you can use a max FWHM of 4 and a min SNR between 5 and 15
- I normally use symmetricity threshold values between 50-80%
The other standard values should work out ok.

The final image of the dataset is not published. You would need a
dataset with much more images to generate a significant reconstruction
with visible structures. The final size of the file would be too big
to make it available online through standard means.

Send me a direct email to paxc...@gmail.com if you want a bigger
dataset and we can try to set it up.

Cheers,
-R

--
Ricardo Henriques, PhD
Institut Pasteur (Paris, France).
For contact information see: https://sites.google.com/site/paxcalpt/

[Jessie Weber]

Hi Ricardo,

Thanks again for your bead suggestion - the beads are working great!

I'm still interested in a 2D dataset as described below - essentially
something that I can fully reconstruct and compare to your final image
- kind of as a standard.

Also, I'm trying to play with the quickpalm output particles tables in
Matlab. I can load an RGB image using imread (green channel only),
which shows the same image as the reconstruction in imagej. However,
I'd like to load the full table and have access to the coordinates.
Although the file is ".xls", there's a formatting when opening in
excel - you have to choose "delimited" - then you can see all the
data. Since the format is not actually ".xls" the xlsread function in
Matlab also fails. Any tips to get the coordinates out of the
Particles Table.xls?

Thanks again very much for your help and I hope your graduation trip
was excellent!

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Ricardo Henriques]

Hi Jessie,

I think the best way to open a QuickPALM particles table in matlab -
at least the way I do it - is to save the particle table into a
tiff-file [QuickPALM->Save Particles (.tiff file - fast)]. This tiff
is not a reconstruction of your data but truly the list of particles
information set into a 32-bit float image, you can then load the tiff
in matlab and convert it into a float matrix, you should have the full
particles list then.

You can see a full description of the .tiff particles table format
here - http://code.google.com/p/quickpalm/wiki/Tutorial_Particle_table

I've had (and have) my hands full for the last months, I'll try to get
you the comparison dataset as soon as I can.

Cheers,
-R

--

[Jessie Weber]

Hi Ricardo,

Once again, thank you for your reply! That should be helpful!

Take your time with the dataset - I've got plenty to do in the
meantime. It's more to prove to collaborators that I can reconstruct
the same data as you with your own algorithm (and also to troubleshoot
any new alogorithms I want to try).

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Jessie Weber]

Hi Ricardo,

I've been playing with the Particles Tables in Matlab and have a couple of hopefully quick questions:

1. I'm seeing inconsistencies between the data imported from the Particles Table.tif and the same table loaded into Matlab. I'm using imread to read in the nx14 data sets (I tried simply load as well, same results). The numbers are the same in the Matlab loaded variable as in the Particles Table.tif loaded in imagej for the first several particles, but at some point, they're no longer the same. Does this sound familiar? Is there something obvious that I'm missing? The only way I can get exactly the same numbers that appear in Particles Table.tif into matlab is by copying and pasting from the .tif file into excel, then loading into Matlab.

2. NAN positions come up in the saved Particles Table.tif that do not show in the Particles Table.xls Could this be due to changes in the .tif after drift correction? The .xls file is saved at the time of analyzing partilces. Is the .tif file updated automatically after drift correction? I am guessing that is the case, since if not there would be no way to access the drift corrected positions...

3. Are the particle intensities reported in the Particles Tables raw counts, or after background subtraction? Are these the values you use to calculate your localization accuracy?

Once again, thank you so much for your code and your help!

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Ricardo Henriques]

Hi Jessie,

1. I'm seeing inconsistencies between the data imported from the Particles Table.tif and the same table loaded into Matlab. I'm using imread to read in the nx14 data sets (I tried simply load as well, same results). The numbers are the same in the Matlab loaded variable as in the Particles Table.tif loaded in imagej for the first several particles, but at some point, they're no longer the same. Does this sound familiar? Is there something obvious that I'm missing? The only way I can get exactly the same numbers that appear in Particles Table.tif into matlab is by copying and pasting from the .tif file into excel, then loading into Matlab.

I actually don't know about this one, I've never tried to import the tables into matlab. Notwithstanding, I frequently convert them into python numpy arrays without issues. There is some small processing that needs to be applied to the pixel info from the "Particles Table.tif" described here http://code.google.com/p/quickpalm/wiki/Tutorial_Particle_table.

2. NAN positions come up in the saved Particles Table.tif that do not show in the Particles Table.xls Could this be due to changes in the .tif after drift correction? The .xls file is saved at the time of analyzing partilces. Is the .tif file updated automatically after drift correction? I am guessing that is the case, since if not there would be no way to access the drift corrected positions..

- Yes, the NaN comes from the drift correction. 

- After drift correction the particles table in RAM memory is updated, if you then save it into a tiff it will have the drift correction applied. If you saved the particles table before drift correction also then you'll be able to have both the drift-corrected and non-drift-corrected particles positions.

3. Are the particle intensities reported in the Particles Tables raw counts, or after background subtraction? Are these the values you use to calculate your localization accuracy?

The particle intensities are given after background subtraction. Yes, I use those values to (roughly) estimate localization accuracy, here's how I do it (extracted from a linkedin comment on the super-resolution group):

"""

... There are several ways to do this, perhaps the easiest is by calculating the theoretical formula from Thompson et al. (Biophys. J. 2002). It indicates that your localization accuracy is similar to the PSF_FWHM / sqrt(N_Photons). If you're using an EM-CCD (that has a linear gain) then N_Photons can be calculated as the division between the particle brightness and the camera gain. But beware, this formula gives you only an approximation and it will over-estimate your localization accuracy. 
Here's an example: 

- my own Andor iXion 897 EM-CCD camera has a measured gain of 37 (if you don't know how to calculate this, I can send you a section of my PhD thesis that explains how) 
- my imaging settings yield PSFs with ~350nm FWHM 
- I've identified a particle with a brightness of 5.000 
- the localization accuracy will be around 350/sqrt(5000/27) = 20nm 
"""

Hope it helps...

Regards,

-Ricardo

[Jessie Weber]

Hi Ricardo,

Thank you very much once again! Could you advise me on the easiest way to get a copy of your thesis? Sounds like a valuable document!

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Ricardo Henriques]

Hi Jessie,

Unfortunately I can't share the whole thesis as there is parts of it being considered for publication. Notwithstanding, here's the small part that discusses how to estimate photon count with a EM-CCD camera: http://goo.gl/ihwFg.

Cheers,

-Ricardo

[Jessie Weber]

Great, thank you! Any idea when it is likely to be available?

Best,
Jessie

--

Jessie Weber,
jessie....@gmail.com

[Ricardo Henriques]

As soon as those papers get out ;).

Probably under a year or so.

[Ye Chen]

Hi, Ricardo,

A related question about the photon counting. As you mentioned in this
email " The particle intensities are given after background
subtraction. "

Is this the intensity in the very center pixel of the PSF, after
background substraction? Or is it an average value over a, say, 3x3 or
5x5 square?

Thanks,

Ye

[Ricardo Henriques]

Hi Ye,

The intensity is calculated by integrating the signal over an area equivalent to the given FWHM, after background subtraction.

The code part where I do this can be found on https://code.google.com/p/quickpalm/source/browse/QuickPALM/MyFunctions.java line 283.

Cheers,

-Ricardo