I completed the experiment for one of our colaborators (I measured inhibition of an enzyme by 3 compounds). Unfortunatelly compounds were not active in the range of dilutions that I had used (up to 10 micromol). Higher concentrations caused the paradoxically high signal due to (probably) interactions with the detection reagent. Therefore concentrations higher than 10 micromol were excluded and, in the end, I could not observe the full dose-dependent curves (and did not calculate IC50). If it was my publication I would left the results like this with the explanation why higher concentrations were not included. But the colaborator asked If I could "complete the curve" on the graph for one of the compounds. I would rather not do that but I consulted one of my supervisors and added dotted line in place where could be the curve (the highest concentration of compound reached ~ 50% of the maximal inhibition, rest of the curve was predicted by the program). I though, that the request was made just to make graph more pleasant to the eye. But later the same collaborator made another request - to calculate IC50 from that curve (where half was from the experiment and half was "predicted" by the program with the assumption that the compound will reach 100% of inhibition - which is not certain). The collaborator also asked to "complete" another curve (where experimental data reached only 25% of the maximal inhibition).
If it was my publication I would never do that. I do not want to suggest that I tested concentrations that I did not test and I do not want to suggest that the compounds were able to fully inhibit enzyme (which I did not prove). Also I do not want to create "artificial IC50" which could be used to compare the compound with different compounds creating false conclusions. But maybe I am wrong? Therefeore I want to ask:
Is it correct practice to "complete the curve" (to elongate the curve to reach 100% of expected effect) even if you did not test higher concentrations (because higher concentrations were to high for the assay). Is it correct to calculate IC50 from that kind of curve?
You can only really make accurate predictions between the IC15 and IC85. As long as you have at least 3 doses within that range you can predict an IC50 using probit analysis. You can in your figures have a predicted curve but you cannot use that for any subsequent analysis.
Robert Adolf Brinzer is right. 15-85 is the part of the sigmoid curve that most resembles a straight line. We can't quantitate reliably outside that region of a calibration curve.
Although we may "predict" results, we must never fabricate data. You might try to publish a curve with points indicating your data and dotted lines suggesting how you think the system may behave at other concentrations. You might get that published in a predatory journal.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology