Re: Densitometry of Western Blots

Posted by Wayne Rasband on Apr 11, 2007; 6:46pm
URL: http://imagej.273.s1.nabble.com/Re-2D-area-calculations-tp3699778p3699780.html

The best solution is to scan the film along with a calibrated OD step
tablet and to generate a calibration curve using the procedure
described at

     http://rsb.info.nih.gov/ij/docs/examples/calibration/

Search the Web for EK1523422 or EK1523406 to find companies that sell
Kodak step tablets.

-wayne

On Apr 10, 2007, at 6:14 PM, Wolfgang Schechinger wrote:

> Dear ImageJ Experts,
>
> I want to quantitate signals on X-Ray films obtained by Western Blot /
> ECL.
> Actually, I just scanned the films with an ordinary flatbed scanner
> (transparency mode, internally 10 bit), that gives me 8bit tiff
> images. I use the Ctrl_1/2/3 sequence from ImageJ to generate density
> graphs of each lane which then are integrated with the wand tool.
>
> Two issues now make me think and I am looking for a good solution:
>
> 1) Strong signals yield spots on the film which probably are bigger
> than the actual spot was on the gel. Longer exposure times make larger
> spots)
>
> 2) I am worried about the meaning of the area under the curve data as
> I have scanned a microplate of a colorimetric assay and compared the
> OD values obtained by the photometer with the grey values of the wells
> obtained by scanning. The relationship is not linear unless I would
> plot the log of the grey values against the OD. That only would make
> sense if one applied Lamber-Beers's law I/Io=10^(-ecd) on the data
> where I is the grey value and ecd is the OD value from the photometer.
>
> I also read in a tutorial on a laser densitometer from Molecular
> Devices that this machine would convert the densitometric values to
> greyscales for storage using a logarithmic relation that would
> resemble the equation mentioned above.
>
> My goal is to compare the amounts of proteins on the blots. May I
> assume (within all the limits the use of xray films has) that double
> the amount of protein yields double the area under the curve or is the
> reality (even under the constraints mentioned) more complex. Is there
> a reason (or even necessity) to use the log of the area in order to
> obtain correct values?
>
> Thanks for your help!
>
> Wo
>