Recently I am doing CHIP-qPCR to test one TF binding. I use 10% of sample as Input and the rest 90% to do IP. I have the TF-positive cell line A and TF-depleted cell line B.
Primer is flanking the targeting site of TF.
The result I have is:
In A: Ct in IP 26 & Ct in Input 20 [Ct difference is 4]
In B: Ct in IP 30 & Ct in Input 20. [Ct difference is 10]
It seems TF binds to the targeting site.
However, I followed the calculation process of other people to estimate yield:
1/9*1/(2^6)
which is extremely low.
I am wondering if anybody has some idea about this: is it reasonable to do calculation like this? What is the average yield you get?
Using delta-delta CT, i have calculated a fold increase for your sample. I used the input value as the reference. If i compared B vs A, i found a fold increase of 16.
There is a difference between calculating binding yield and relative enrichment, although the terms are used interchangeably.
Relative enrichment is for contrasting samples, and is otherwise known as a delta-delta Ct (see caveat below), and measures the difference between samples, irrespective of the global efficiency of your pull-down, which is antibody- and buffer-dependent (more on that below). In your case, it is indeed 2^[(30-20)-(26-20)] = 2^4 = 16, as Stephane calculated, since your dilution factors all cancel out in the ratio.
The yield calculation is for individual samples, which is what you did, and it takes into account the fact that you used 90% of your samples for the IP and the difference between input and ChIP, and thus your yield is very low - but not atypically for ChIP. This number is dependent on TF occupancy, efficiency of crosslinking and its reversal, efficiency of immunoprecipitation, stringency of washes etc.
If you do the yield calculation for each sample and then work out the ratio between them, you'll be back to 2^4.
Importantly, the caveat for all Ct-based calculations is that it assumes your qPCR efficiency is 100% - do you have a standard curve for it? Current "miQE guidelines" (search term in PubMed) for proper qPCR experiments pretty much makes a curve compulsory if you are making any substantial quantitative statements.
As a last comment, you have to include a qPCR for an unrelated locus to show that the loss of your binding site is specific, so essentially to account for differential non-specific binding between the samples (e.g. better washing in one tube).
Good luck!
Hello Khezhuo
I will just make a few comments on your experimental design, as I feel like the provided information is insufficient.
From what I understood you are comparing two samples TF+ and TFnull.
From these samples you will need 3 elements, the input (i), the mock (to determine the background pull down of a random antibody) (nc) and the anti-TF antibody (tr).
Regarding the targeting you will need a pair of primers that target the TF binding site (bs) and a control primer in a region that is not bound by your TF (negctrl) and that is at a greater distance than your sonication fragment size.
Once you have all these elements and data you can proceed to calculations:
I will assume that you are doing a dilution factor of your input of 10 so you will have to have a correction factor of log(1/10)=3.321928
1-delta cts and error associated
dCT.tr = CT.i - (CT.tr-3.321928)
dCT.nc = CT.i - (CT.nc--3.321928)
dSD.tr = sqrt( (SD.i)^2 + (SD.tr)^2 ) / sqrt(n)
dSD.nc = sqrt( (SD.i)^2 + (SD.nc)^2 ) / sqrt(n)
2-delta delta cts and error associated
ddCT = dCT.tr - ddCT.nc
ddSD = sqrt( (dSD.tr)^2 + (dSD.nc)^2 )
3-fold change and error associated
FC = 2^(ddCT)
FC.error = ln(2) * ddSD * FC
- Badges
- Science topic