ELISA vs. Western Blotting - Information on repeatability and resolution of these 2 techs? (Sep/28/2008 )
I have been trying to assess the best technique for assaying target protein levels among plant populations. I need to know which technique will yield the best accuracy and resolution:
Western Blotting produces great gel-to-gel variation and also within-gel variation even when I use a single sample in every well of the gel.
ELISA is said to be able to produce exactly quantitative measurements of protein concentration.
So first, is that true?
What is the repeatability of ELISA? In other words, if I do 4 replicates of the same sample in ELISA will I see significant variation in results?
Is there formal documentation of these differences? Or just of the accuracy of either Western Blot or ELISA?
I don't necessarily need to know the exact ng/ml concentration of my target protein. I just need to know that in either technique, if two samples show a 10% difference, then it is really a 10% difference in concentration and not just variability due to gel-to-gel/within-gel variation or whatever similar factors might effect ELISA.
Generally ELISA will give more reliable data, but also depends on antibody specificity and sensitivity.
That makes good sense, but the question is, is there a way to determine if that is true without buying all of the reagents necessary?
I think ELISA, if well set up, is more precise. You can obtain very low intra and also inter assay variation, up to 5%. Western can be used for quantitative measurements, but in my own experience results are always less precise and accurate and the procedures is much longer and needs more steps, each of them potentially introducing variability factors
Lorrx: Just to clarify, you mean that any given sample, if assayed via ELISA several times will on average be only 5% different from the other replicates of that sample?
That, I think, is pretty good!
intra e inter assay CVs are usually considered as the ratio between the standard deviation of the analysed samples and their mean (multyplied 100). It means that if you have a 5% intra-assay CV your samples' values are, on the average, 5% different from their mean. But the lower and the higher value diverge more than 5%, of course. However, normally, intraassay e interassay CVs are respectively <10% and <20%
Great explanation of Coefficient of Variation. Is there data on this that is available somewhere? Something I could cite? I can't find anything. The only option otherwise for me would be to actually buy all the necessary reagents and test it myself... clearly an unacceptable proposition.
I know about this percentages cause I have some personal experience and read lot of papers, but they simply present results giving explicit values of CVs. nothing on the pure theory you could cite. Try to search something on the web on intra-interassay CV calculation theory. There should be something somewhere. Once i had a terrific book on this statistics, but unfortunately it remained in my ex lab...I'm sorry. I think that the ELISA way worths a trisl. Otherwise you could try with quantitative PCR to measure the mRNA related to the target protein you are looking for, but, of course, it gives not exactly the same kind of information...