I am working on drug target identification using an overexpression (OE) Library. The approach relies on the calculation of drug EC50 shift among OE cell lines. I am dealing with the issue that in some cases, the OE seems to be toxic to the cells, affecting the cell growth rate.
The workflow used was based on the normalisation of the data considering positive and negative control points and posterior analysis using GraphPad Prism to calculate the EC50 using nonlinear regression (log Agonist vs normalised response).
Once I have normalised the data considering controls, the OE cell line shows a top value about 160-200% of survival (Somehow, in low concentrations, the drug-target interaction actually blocks the toxicity of the target OE and then these points have more cells rather the control (just OE cells), and subsequently the percentage of survival is greater at these points.
GraphPad gives the option of fitting the dose-response curve normalising the data again, constraining the curve to run from 0 to 100%.
The problem is that with the double normalisation the results are different if I compare it with a single normalisation considering controls, and not fitting the curve again for a range 0-100. Thus, I wonder how many times should I normalise the data to adapt the slope of the curve facing the different behaviour of the points according to cell growth.
I'm a bit confused. If you calculate an EC50 shift assay, in my mind, it is normal that the percentage of survival shows greater than control in low conc. So do you mean IC50?
Anyway, I think you should take double normalisation. We perform a normalisation compared to control to reduce the difference between repeats. And a normalisation of fitting curve is used to make the curve more close to the fact.
So when i perform a IC50 assay, I always normalise the data twice. First change the activity to DMSO%. Second set the constrain parameter as in the picture, for the top and bottom of activity (you can also choose inhibitition and the EC50 assay shift assay equation) with inhibitor should be 100 and 0 in the abstract.
Hi!
I would recommend to set up your normalization and later analysis using a couple of tool compounds as positive and negative controls. I.e., a regular cell inhibitor which mechanism of action is not involved with your overexpression target would not be likely shifting your EC50 values neither the curve slope; the other way round, using a pan-killer of your desired target would give you the expected shift.
Once you have established the proper way to analyze your data, then you can proceed your analysis with confidence.
All the best!
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Molecular Biological Techniques