I guess the verdict is still out exactly how well the Qubit Protein Assay works. Now that I’m looking into the details, looks like the Qubit Protein Assay (Q33211) that we have is incompatible with RIPA buffer, which is likely the buffer people had tried using it with in the past. But, the BCA assay is perfectly compatible with RIPA buffer, so we still do it reasonably often. But, to make sure everyone in the lab is on the same page, we may as well look at some real data and see how I view it. This will be based on some data that Olivia S had generated a month or so ago.

Link to the code below:
https://colab.research.google.com/drive/16jIz0Qq_jvb7R1s-fIJW804ffahZDHPG?usp=sharing

Here’s what the raw values of the standard curve looked like:

The value of the background is shown as the datapoint on the y-axis. Clearly, the lowest standard dilution loses linearity from the rest of the values, likely b/c a larger fraction of it is due to background signal. Accordingly, if we subtract out the background value from it, we might be able to salvage use of that dilution. That is indeed the case.

Using those values, I’ve now fit a linear model. This will allow us to predict the concentration of a protein sample based on its absorbance, relative to the standard curve. The points and line in red denote the values calculated based on this linear model. The real standard curve values do straddle it pretty well, which is good.

Cool, so what do the experimental samples of unknown concentration look like based on this model? Since Olivia did half-log dilutions of these, we can actually see how the predicted values change based on the dilution tested and its resulting absorbance value.

So for all of the samples, the 33-fold dilution sample was erroneous. Otherwise, 10 uL of undiluted sample in 190 uL BCA working solution, or 10 uL of 3- or 10-fold dilutions of that lysate in 190 uL BCA working solution, gave values that were relatively similar. Taking the geometric mean of those, we calculated values of 1184, 1894, and 2339 ng / uL for these lysates. Seems reasonable enough.

There we go. A primer on the data analysis portion of the BCA assay, or anything else that requires using dilutions of a standard of known concentration to determine the likely concentrations of unknown samples.