This is a big problem because more and more journals are asking for quantitation on western blotting. In a perfect world where we have limitless cash for our work we would all do ELISAs. The reason I say this is that with an ELISA we always make standard curves to make sure we are taking our data points in the linear range. Western blots are a bit different because we have no standard curve we have to try do the experiment such that we give ourselves the best chance of our bands being in that linear range. We need to think about the actual mechanics of the experiment-we are looking at 3 different reactions-the target with the primary, the primary with the secondary and then the ECL reaction. Sheona is absolutely right about ECL and band density and that Odyssey or another fluorescent technique is better, and Nicolas makes the really good point that you could run a series of dilutions of your sample, and construct a curve and then you would be able to measure the gradient and compare the gradients (in the linear range) between two different samples.

1 If we use an automatic developer then we should do a number of exposures and always take an underexpossed X-Ray film. If we take a saturated film then there may well be a difference in the protein levels in reality, but because the film is saturated then the scanning and quantitation will not detect it.

2 If we have to develop our X-Ray films manually then this adds another layer of complication: Imagine a situation where you have saturated your x-ray film exposure during the ECL step, but then develop the film such that it appears by eye not be saturated, we could be fooling ourselves that the actual x-ray is fine which it isn't.

3 We all take multiple exposures of x-ray films, but we should also bear in mind that the ECL is a reaction and the reactants covering the band will be used up more quickly in a high expression band than a low expression band after your Nth exposure I have done immunoprecipitations where I can just see the band from heavy chain in the dark room after ECL, but after 2 minutes all the ECL has been used up and you actually start seeing a footprint. If you know you want to scan the film do a series of exposures to the point that you feel that the exposure is right, then re-ECL the filter for that time.

4 Finally Eyeball 1.0 is probably the best densitometer out there! if you cannot see a difference by eye then it probably is not there!

Hi Mohamed Alalem

Try Bio rad's Versa doc/ chemi doc instument to image your X-ray/ PVDF membrane and use Image lab software. This software is user friendly, where it will select the band from the membrane automatically when you place the band select tool on band. Even you can define and set the range of you basal expression band and it will quantify automatically intensity and express you results with respect to that of basal expression.

All the best

Thank you very much Sundarajan Kannan for your kind care and assistance.

Two step answer:

1/ I prefer to scan my film, which never reach saturation (always keep a grey scale) using positive lightening of film scanning at 300 dPi, and using black and white encoding to avoid artifact of coloring. I save them in tiff 16bit to keep format opened.

http://www.colorneg.com/scanning_slides_and_negatives/scans/Epson/Perfection/Epson_Scan/

Then I open them with imageJ, apply a rolling ball correction and select a rectangular band to keep a background stable over all my bands of interests.

Keep in mind that this quantification is relative to the highest column of black pixel, so you can not compare directly two blots using this method!

2/ Any way, quantifying blot on film is indicative but such far to be a really quantification... They are too many parameters not linear: luminescence signal, film printing, scanning...

http://www.licor.com/bio/QuantitativeWesterns/pdf/bt0609_LI-COR_final.pdf

The only way to make it not to inaccurate is to use several dilutions of the same samples to confirm your results:

http://www.ncbi.nlm.nih.gov/pubmed/17482639

If you have access to it, GE's ImageQuant software gives very good results.

One thing to bear in mind is that the relationship between band intensity (measured in pixel intensity if you are using ImageJ or similar) and protein concentration is NOT linear when using ECL reagents. A better, and linear, alternative is to use fluorescent secondary antibodies and scan the blot using either a Typhoon (GE Healthcare) or, with the correct fluorescent secondaries, using the Odyssey (Licor) system for example. Both of these systems have internal software for this purpose.

If this in not possible, I agree with the suggestion above that you will need to load several dilutions of your protein mixture and ideally each of these should be loaded in triplicate to allow averaging for loading and transfer variability.

Hope this helps.

I have tried several softwares for the signal density. Yes, there will be discrepancy among the readings but after correcting aganist the house keeping protein (loading control, or whatever you call it) the ratio should be very similar. But if you adjust the sofware, say contrast range (linear, sigmoid, or skewed) then the readout will change. ImageJ, a free sofware, has its reputation since most researchers use it, should be god for the purpose. One thing need to keep in mind is lng exposure can easily take away the difference since all signals will be saturated.

I have to apologize that I meant ImageJ should be "good" not "god" for the purpose. My keyboard is too old to respond, especially the key "o".

So you can just edit your previous post using "edit" close to the "rate up/down" :)

I'll add that there's a tutorial on how to do it with ImageJ here:

http://howtowesternblot.net/data-analysis-3/quantification/

There is a good publication on this: http://www.ncbi.nlm.nih.gov/pubmed/19517440

Good day Mohamed Alalem,

From my experience, Quantity One from Biorad can be used to analyze protein band density (if you can get access to one). I never use ImageJ though. For this software, you can scan you image but save as TIFF format instead of JPEG.

All the best.

There is another free software for analyzing Western blots available from LI-COR. You can download it at http://www.licor.com/islite. It automatically applies a localized background subtraction, which helps reduce some of that subjective variability. It is simple to use and also has a management database to keep track of all your data. You could try that out.

This is a big problem because more and more journals are asking for quantitation on western blotting. In a perfect world where we have limitless cash for our work we would all do ELISAs. The reason I say this is that with an ELISA we always make standard curves to make sure we are taking our data points in the linear range. Western blots are a bit different because we have no standard curve we have to try do the experiment such that we give ourselves the best chance of our bands being in that linear range. We need to think about the actual mechanics of the experiment-we are looking at 3 different reactions-the target with the primary, the primary with the secondary and then the ECL reaction. Sheona is absolutely right about ECL and band density and that Odyssey or another fluorescent technique is better, and Nicolas makes the really good point that you could run a series of dilutions of your sample, and construct a curve and then you would be able to measure the gradient and compare the gradients (in the linear range) between two different samples.

1 If we use an automatic developer then we should do a number of exposures and always take an underexpossed X-Ray film. If we take a saturated film then there may well be a difference in the protein levels in reality, but because the film is saturated then the scanning and quantitation will not detect it.

2 If we have to develop our X-Ray films manually then this adds another layer of complication: Imagine a situation where you have saturated your x-ray film exposure during the ECL step, but then develop the film such that it appears by eye not be saturated, we could be fooling ourselves that the actual x-ray is fine which it isn't.

3 We all take multiple exposures of x-ray films, but we should also bear in mind that the ECL is a reaction and the reactants covering the band will be used up more quickly in a high expression band than a low expression band after your Nth exposure I have done immunoprecipitations where I can just see the band from heavy chain in the dark room after ECL, but after 2 minutes all the ECL has been used up and you actually start seeing a footprint. If you know you want to scan the film do a series of exposures to the point that you feel that the exposure is right, then re-ECL the filter for that time.

4 Finally Eyeball 1.0 is probably the best densitometer out there! if you cannot see a difference by eye then it probably is not there!

I bought the Odyssey for my lab and will never do an ECL reaction again.

The problem is the inherent non-linearity of the film. Pre-flashing can remove low signal anomalies but you still have to deal with saturation issues at the high end. In contrast imaging systems provide much more linear results and should improve the quality and reproducability of the data.

Paul is right. There is a detailed paper that discusses this on the LI-COR website. The link is attached below.

http://biosupport.licor.com/docs/WhitePaper_FilmPhotochemistryEffects_v4_979-13806.pdf

I agree with M. Scheid, peroxydase enzymatic reaction give much more artifacts as the fluorescence secondary antibodies.

More precisely, in top of saturation effect or even "burned signal", we also observed in our lab that enzymatic detection are also less reproducible. Same membrane detected twice do not give always exactly the same signal.

BUT... fluorescent detections are less sensitive... so luminescence still make sens in difficult conditions.