- If you have the actual assay validation data, that is best. You can input the standard concentrations and the inter- or intra- day standard deviation in to the Pmetrics makeErrorPoly() function and choose the best set of coefficients.
- If you don't' have the assay validation data, my usual next step is to set C0 = 0.5 * LOQ and C1 = 0.10 and then allow gamma to be fitted.
- If I'm getting poor results, I will try the same, but use lambda instead of gamma, and allow lambda to be fitted.
- You can also try a constant assay error where C0 !=0 and C1:C3 = 0.
- You can allow Pmetrics to estimate the coefficients based on the data. Recognize that the estimates will likely vary depending on your model structure and parameter ranges. To do this look at the help file for ERRrun in Pmetrics. ERRrun is a special implementation of IT2B which estimates the assay error polynomial coefficients. Just like for IT2B, you need a model file and data file in your working directory. The first time you do the run, without an instruction file, you will answer all the questions as you would for IT2B, except when you get to the section on the assay error. When asked, you will answer that you have no idea about the coefficient values and that you would like the program to estimate them for you. The output of this run will be a file "ASS0001" whose last line will contain the estimated values for C0:C3.
Post technical questions about making Pmetrics carry out a particular task.
1 post • Page 1 of 1
To experiment with the error model, especially if some concentrations (e.g. peaks) are being consistently poorly predicted, try the following.