WB Analysis Convert Histogram area to OD

[MKR]

I feel like I am constantly searching for just the right way to analyze Western Blots.

I like the CTRL-1 , 2, 3 Gel analysis method which brings up a histogram of each selected lane. Then I can see the background, draw a line so that I know I am getting just the band of interest. However, what is that number?

If I am uncalibrated, I'm assuming it's a pixel gray value, but an area (integration?) of the pixel gray values in the band of interest.

If I calibrate to OD or do uncalibrated OD, I'm assuming it's an area of the OD values of the band.

If I take several bands and get values of these areas (both methods), then normalize to the highest (or darkest, strongest band) by setting it to 100, then both methods give different results. Obviously they give the same order of bands (as in which band is the lowest, 2nd,...), but the ratios are different.

If I do the ROI method for OD, I get another ratio (and yet another ratio if I choose different ROI sizes, shapes, etc.)


QUESTION
So with the ROI method, you can get an OD value and report or label your graph as OD units. What do you report for the OD areas or pixel gray value areas that you get?  Can I 'convert' (or normalize) the numbers to OD?
QUESTION


I want to do a WB with standards at different known dilutions so I can see which method of analysis gives a more real life ratio. Has anyone done such a thing?

Thanks for any input.

[Johannes-P. Koch]

There is quite a good tutorial available here:
http://www.lukemiller.org/journal/2007/08/quantifying-western-blots-without.html ;
just skip the part concerning Adobe Photoshop and read the section about
ImageJ, there you get several methods how to quantify.

best regards

_________________________________
Johannes-P. KOCH
University of Vienna
MFPL, Center of Molecular Biology
Department of Biochemistry
Dr. Bohrg. 9/5
A-1030 Vienna
Austria
phone [phone]
fax [phone]

[hidden email]



----- Original Message -----
From: "MKR" <[hidden email]>
To: <[hidden email]>
Sent: Wednesday, January 14, 2009 11:38 PM
Subject: WB Analysis Convert Histogram area to OD

>I feel like I am constantly searching for just the right way to analyze
> Western Blots.
>
> I like the CTRL-1 , 2, 3 Gel analysis method which brings up a histogram
> of
> each selected lane. Then I can see the background, draw a line so that I
> know I am getting just the band of interest. However, what is that number?
>
> If I am uncalibrated, I'm assuming it's a pixel gray value, but an area
> (integration?) of the pixel gray values in the band of interest.
>
> If I calibrate to OD or do uncalibrated OD, I'm assuming it's an area of
> the
> OD values of the band.
>
> If I take several bands and get values of these areas (both methods), then
> normalize to the highest (or darkest, strongest band) by setting it to
> 100,
> then both methods give different results. Obviously they give the same
> order
> of bands (as in which band is the lowest, 2nd,...), but the ratios are
> different.
>
> If I do the ROI method for OD, I get another ratio (and yet another ratio
> if
> I choose different ROI sizes, shapes, etc.)
>
>
> QUESTION
> So with the ROI method, you can get an OD value and report or label your
> graph as OD units. What do you report for the OD areas or pixel gray value
> areas that you get?  Can I 'convert' (or normalize) the numbers to OD?
> QUESTION
>
>
> I want to do a WB with standards at different known dilutions so I can see
> which method of analysis gives a more real life ratio. Has anyone done
> such
> a thing?
>
> Thanks for any input.
> --
> View this message in context:
> http://n2.nabble.com/WB-Analysis-Convert-Histogram-area-to-OD-tp2159684p2159684.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
>

[MKR]

Not to be ungrateful, but I'm not looking for a way to quantify!?

In fact my original post goes over many ways to quantify. Thanks for the link though (I had seen it in the past).

My question is about 'converting to OD' and how to 'report' or compare area under the histogram curves. The link provided talks about (in the third method) doing the histogram curve area - saying 'There will likely be very little difference in the results between the various methods.' And in a way this is true - but it depends on your definition of very little difference. I definitely see differences in each method. Tomorrow I will run some known samples in two dilutions and see which method gives the expected ratio when quantifying.

Johannes-P. Koch wrote

There is quite a good tutorial available here:
http://www.lukemiller.org/journal/2007/08/quantifying-western-blots-without.html ;
just skip the part concerning Adobe Photoshop and read the section about
ImageJ, there you get several methods how to quantify.

best regards

_________________________________
Johannes-P. KOCH
University of Vienna
MFPL, Center of Molecular Biology
Department of Biochemistry
Dr. Bohrg. 9/5
A-1030 Vienna
Austria
phone [phone]
fax [phone]

[e-mail]



----- Original Message -----
From: "MKR" <[e-mail]>
To: <[e-mail]>
Sent: Wednesday, January 14, 2009 11:38 PM
Subject: WB Analysis Convert Histogram area to OD


>I feel like I am constantly searching for just the right way to analyze
> Western Blots.
>
> I like the CTRL-1 , 2, 3 Gel analysis method which brings up a histogram
> of
> each selected lane. Then I can see the background, draw a line so that I
> know I am getting just the band of interest. However, what is that number?
>
> If I am uncalibrated, I'm assuming it's a pixel gray value, but an area
> (integration?) of the pixel gray values in the band of interest.
>
> If I calibrate to OD or do uncalibrated OD, I'm assuming it's an area of
> the
> OD values of the band.
>
> If I take several bands and get values of these areas (both methods), then
> normalize to the highest (or darkest, strongest band) by setting it to
> 100,
> then both methods give different results. Obviously they give the same
> order
> of bands (as in which band is the lowest, 2nd,...), but the ratios are
> different.
>
> If I do the ROI method for OD, I get another ratio (and yet another ratio
> if
> I choose different ROI sizes, shapes, etc.)
>
>
> QUESTION
> So with the ROI method, you can get an OD value and report or label your
> graph as OD units. What do you report for the OD areas or pixel gray value
> areas that you get?  Can I 'convert' (or normalize) the numbers to OD?
> QUESTION
>
>
> I want to do a WB with standards at different known dilutions so I can see
> which method of analysis gives a more real life ratio. Has anyone done
> such
> a thing?
>
> Thanks for any input.
> --
> View this message in context:
> http://n2.nabble.com/WB-Analysis-Convert-Histogram-area-to-OD-tp2159684p2159684.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
>

[Jonathan Hilmer]

> I like the CTRL-1 , 2, 3 Gel analysis method which brings up a histogram of
> each selected lane. Then I can see the background, draw a line so that I
> know I am getting just the band of interest. However, what is that number?

I assume the number you're referring to is the automated measurement
you get with the area selection tool.  That measurement is literally
just the area of the selection, exactly as you see it.  There is some
normalization/scaling code that runs in the course of creating the
plots, so the dimensions and measurements from the lane plots do NOT
correspond directly to any kind of calibrated units.  However, they
should be linearly scaled and measurements within one plot-image
maintain appropriate relative sizes.  It's been a while since I looked
at the source code, but I think the only exception to this would be
for very small images which are not scaled down.  I doubt you'd have
plots that small, however.

It is possible to alter the code to display the scaling factors to the
user (which you would then use to correct the measurements manually
after quantitation), or remove the scaling (which generates huge plot
images) so that the area measurements reflect the actual integrated
density, but that's not part of the standard GelAnalyzer.

You can not directly quantitate -between- multiple runs of GelAnalyzer
plots with different selections, because this scaling will probably be
different for each run.  Relative quantitation needs to be all within
one ctrl-1,2,2,2,etc,3 set.

Some of the other tools, such as plot-profile and the general
measurements, are better suited to make direct measurements of the OD,
since they don't scale the numbers before presenting them to you.  The
tradeoff is that you don't have the interactive definition of
background and ROI as you do with the GelAnalyzer plots.

My personal preference has been to use modified GelAnalyzer code to
generate large plots, but without any correction to actual OD units.
The relative intensities give me the measurments I need.  However,
this is for gels with features (bands) that run very close together
and with varying background, so the definitions of band borders and
the background are very important.  If your blots have well-separated
features with low background then it would be better to just use the
measurements tool after specifying appropriate rectangular or
user-defined ROIs.

Jonathan