background subtraction

[Knecht, David]

We are trying to determine the distribution of a GFP-fusion protein in a cell. All the images are 16 bit and confocal. If one probes with GFP alone, there are areas of the cell that appear significantly brighter than other areas because they are vesicle free (so called hyaline cytoplasm) and so the concentration of free cytoplasmic proteins is higher.  Our fusion protein probe localizes to one of these vesicle free regions.  So if you double label a cell with GFP-fusion protein and RFP, the same areas appear higher than background, but the ratio of hyaline localized to normal cytoplasm is higher for the fusion protein than the FP alone.  Therefore, we want to use the unfused protein as "background" and subtract that signal from the fusion protein signal in order to correct for the non-uniform background. My plan is to set the non-localized cytoplasmic area of the cell to similar intensity values and then subtract with the Image Calculator.  So lets say the RFP image values range from 50 to 1000 and the GFP values range from 200-2000. I am unsure how to correct the images before subtraction.  You cannot "apply" a look up table to the 16 bit images to correct for the two images not being collected at that same intensity level.  What is the most appropriate way to "adjust" the intensity levels of a 16 bit image to equalize two images?  THanks- Dave

Dr. David Knecht    
Department of Molecular and Cell Biology
Co-head Flow Cytometry and Confocal Microscopy Facility
U-3125
91 N. Eagleville Rd.
University of Connecticut
Storrs, CT 06269
860-486-2200
860-486-4331 (fax)

[Aryeh Weiss]

I doubt that there is

On 2/8/11 5:44 AM, David Knecht charter wrote:

> We are trying to determine the distribution of a GFP-fusion protein
> in a cell. All the images are 16 bit and confocal. If one probes with
> GFP alone, there are areas of the cell that appear significantly
> brighter than other areas because they are vesicle free (so called
> hyaline cytoplasm) and so the concentration of free cytoplasmic
> proteins is higher.  Our fusion protein probe localizes to one of
> these vesicle free regions.  So if you double label a cell with
> GFP-fusion protein and RFP, the same areas appear higher than
> background, but the ratio of hyaline localized to normal cytoplasm is
> higher for the fusion protein than the FP alone.  Therefore, we want
> to use the unfused protein as "background" and subtract that signal
> from the fusion protein signal in order to correct for the
> non-uniform background. My plan is to set the non-localized
> cytoplasmic area of the cell to similar intensity values and then
> subtract with the Image Calculator.  So lets say the RFP image values
> range from 50 to 1000 and the GFP values range from 200-2000. I am
> unsure how to correct the images before subtraction.  You cannot
> "apply" a look up table to the 16 bit images to correct for the two
> images not being collected at that same intensity level.  What is the
> most appropriate way to "adjust" the intensity levels of a 16 bit
> image to equalize two images?  THanks- Dave
>

I doubt that there is any validity to comparing fluorescence of two
different FPs measured in two different channels. However, it might be
appropriate  to normalize the  RFP to its value in the vesicle free
area, and this may provide a multiplicative correction factor for your
GFP fusion.

--aryeh
--
Aryeh Weiss
School of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: [phone]

[Knecht, David]

Hi Aryeh- I don't understand your fundamental concern.  We compare patterns of fluorescence in different channels all the time.  How is this different from comparing two channels in a co-localization analysis? It is the relative amount of signal in different parts of the same cell that is in question. Dave

On Feb 8, 2011, at 1:38 AM, Aryeh Weiss wrote:

> I doubt that there is
>
> On 2/8/11 5:44 AM, David Knecht charter wrote:
>> We are trying to determine the distribution of a GFP-fusion protein
>> in a cell. All the images are 16 bit and confocal. If one probes with
>> GFP alone, there are areas of the cell that appear significantly
>> brighter than other areas because they are vesicle free (so called
>> hyaline cytoplasm) and so the concentration of free cytoplasmic
>> proteins is higher.  Our fusion protein probe localizes to one of
>> these vesicle free regions.  So if you double label a cell with
>> GFP-fusion protein and RFP, the same areas appear higher than
>> background, but the ratio of hyaline localized to normal cytoplasm is
>> higher for the fusion protein than the FP alone.  Therefore, we want
>> to use the unfused protein as "background" and subtract that signal
>> from the fusion protein signal in order to correct for the
>> non-uniform background. My plan is to set the non-localized
>> cytoplasmic area of the cell to similar intensity values and then
>> subtract with the Image Calculator.  So lets say the RFP image values
>> range from 50 to 1000 and the GFP values range from 200-2000. I am
>> unsure how to correct the images before subtraction.  You cannot
>> "apply" a look up table to the 16 bit images to correct for the two
>> images not being collected at that same intensity level.  What is the
>> most appropriate way to "adjust" the intensity levels of a 16 bit
>> image to equalize two images?  THanks- Dave
>>
>
> I doubt that there is any validity to comparing fluorescence of two
> different FPs measured in two different channels. However, it might be
> appropriate  to normalize the  RFP to its value in the vesicle free
> area, and this may provide a multiplicative correction factor for your
> GFP fusion.
>
> --aryeh
> --
> Aryeh Weiss
> School of Engineering
> Bar Ilan University
> Ramat Gan 52900 Israel
>
> Ph:  972-3-5317638
> FAX: [phone]

Dr. David Knecht    
Department of Molecular and Cell Biology
Co-head Flow Cytometry and Confocal Microscopy Facility
U-3125
91 N. Eagleville Rd.
University of Connecticut
Storrs, CT 06269
860-486-2200
860-486-4331 (fax)

[Aryeh Weiss]

Hi David,

In colocalization analysis you are comparing morphological features. The
intensities enter into the correlation  (which I think is not always the
right thing to do), but basically you are looking at location.

Here, If I understand your problem correctly, you are trying to
determine how the fluorescence of your fusion protein would have
increased even if there were no vesicles. If you wish to compare this
to a different FP, in a different channel with different gain, then it
appears to me that the correct comparison is relative change (ie, the
free FP changed  by some factor, so that is what the fusion would do if
it were not entering the vesicles).

It is possible I misunderstood the problem, in which case I apologize
for the confusion.

Best regards
--aryeh


On 2/9/11 2:37 PM, David Knecht wrote:

> Hi Aryeh- I don't understand your fundamental concern.  We compare
> patterns of fluorescence in different channels all the time.  How is
> this different from comparing two channels in a co-localization
> analysis? It is the relative amount of signal in different parts of
> the same cell that is in question. Dave
>
> On Feb 8, 2011, at 1:38 AM, Aryeh Weiss wrote:
>
>> I doubt that there is
>>
>> On 2/8/11 5:44 AM, David Knecht charter wrote:
>>> We are trying to determine the distribution of a GFP-fusion
>>> protein in a cell. All the images are 16 bit and confocal. If one
>>> probes with GFP alone, there are areas of the cell that appear
>>> significantly brighter than other areas because they are vesicle
>>> free (so called hyaline cytoplasm) and so the concentration of
>>> free cytoplasmic proteins is higher.  Our fusion protein probe
>>> localizes to one of these vesicle free regions.  So if you double
>>> label a cell with GFP-fusion protein and RFP, the same areas
>>> appear higher than background, but the ratio of hyaline localized
>>> to normal cytoplasm is higher for the fusion protein than the FP
>>> alone.  Therefore, we want to use the unfused protein as
>>> "background" and subtract that signal from the fusion protein
>>> signal in order to correct for the non-uniform background. My
>>> plan is to set the non-localized cytoplasmic area of the cell to
>>> similar intensity values and then subtract with the Image
>>> Calculator.  So lets say the RFP image values range from 50 to
>>> 1000 and the GFP values range from 200-2000. I am unsure how to
>>> correct the images before subtraction.  You cannot "apply" a look
>>> up table to the 16 bit images to correct for the two images not
>>> being collected at that same intensity level.  What is the most
>>> appropriate way to "adjust" the intensity levels of a 16 bit
>>> image to equalize two images?  THanks- Dave
>>>
>>
>> I doubt that there is any validity to comparing fluorescence of
>> two different FPs measured in two different channels. However, it
>> might be appropriate  to normalize the  RFP to its value in the
>> vesicle free area, and this may provide a multiplicative correction
>> factor for your GFP fusion.
>>
>> --aryeh -- Aryeh Weiss School of Engineering Bar Ilan University
>> Ramat Gan 52900 Israel
>>
>> Ph:  972-3-5317638 FAX: [phone]
>
> Dr. David Knecht Department of Molecular and Cell Biology Co-head
> Flow Cytometry and Confocal Microscopy Facility U-3125 91 N.
> Eagleville Rd. University of Connecticut Storrs, CT 06269
> 860-486-2200 860-486-4331 (fax)
>

--
Aryeh Weiss
School of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: [phone]