I am currently working as an engineer and I run large columns in a pharmaceutical factory. I want to extract the file in Unicorn in excel format and use it for mechanistic modeling. The problem is that I do not have real-time concentration data based on UV, but only the total concentration of the eluted solution. Since it is a large column, it is also very expensive to run the experiment. Therefore, I came up with the following idea

Assumption: UV levels are linearly correlated with protein concentration.

get the elution profile integrated over the volume

divide that integral by the total protein quantity, which is eluted from the column, to get the proportionality constant C.

multiply the UV profile by the proportionality constant C to get a concentration-volume graph

Just a quick clarification, if you donâ€™t have real-time concentration data based on UV, then where is the elution profile you mentioned in point 1 coming from?

What I would do is a calibration curve for your detector: bypass the columns and inject a series of your samples of different known concentrations and volumes into the system and record the UV profile. Integrate the peak to get the area. Since there is a series of samples, plot all of them vs. known masses and get the slope C. Ideally itâ€™s a simple function like y=Cx, which indicates your assumption is also correct. From this experiment you would also know the cutoff value (saturation) for your UV detector and know the range where the linearity assumption fails. After this, you can divide the UV profile by C to convert it to a concentration profile.

I also agree with this approach, but it is important to note that UV is not an entirely linear function of concentration. From all my AKTA work, it is linear up to 1500 mAU and then becomes increasingly nonlinear until saturation. Therefore, the fitting to bypass injections should be with a second order polynomial, not a linear fit.

Thanks for the comprehensive and insightful responses both of you have provided.
Can I ask another question? In academia, is it generally accepted to consider the relationship between UV (in Au units) and concentration as a second-order polynomial across all ranges?

Sure thing, glad to help. You can reference my paper in Journal of Chromatography A or Vijesh Kumarâ€™s paper in the same journal, respectively (see below).

â€śIsotherm model discrimination for multimodal chromatography using mechanistic models derived from high-throughput batch isotherm dataâ€ť

â€śRobust mechanistic modeling of protein ion-exchange chromatographyâ€ť

I followed the method to convert UV to concentration outlined by Vijesh. I think more people assume a linear relationship with a saturation cutoff, but this is problematic if you work the nonlinear detection range. It is well known that linear detection ranges are present in uv spectroscopy. Alternatively, you can use different wavelengths that yield lower signal (e.g. 295 nm instead of 280 nm) so that your signals are linear.