Hi all, this is my first time posting here. Thank you for reading this now and being willing to help!

I have been using the nec (NormExp Background Correction and Normalization Using Control Probes) from limma before the actual normalization step for my proteomics antibody array data from RayBio.
(Page 134 in chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.bioconductor.org/packages/devel/bioc/manuals/limma/man/limma.pdf)

Most of the time, the values for negative control based on the intensity signal are lower than that signal for any proteins, which makes sense. So I applied the following code after making sure the data is appropriately organized and aligned.

sample_nec <- nec(samplel_ElistRaw, negctrl="negative", offset = OFFSET) # set the offset to 18, robust = FALSE

However, after I do this, the corrected values tend to have consistent digits after the decimal, for instance,

Before correction
     Sample 1    Status
[1,] 61          negative
[2,] 57          negative
[3,] 82          protein
[4,] 95          protein
[5,] 88          protein

After correction
     Sample 1       Status
[1,] 23.01129     negative
[2,] 21.60262     negative
[3,] 42.72727     protein
[4,] 55.72727     protein
[5,] 48.72727     protein

which I figured out is REASONABLE because the function involved a key formula: raw signal - mu of negative control - (sigma^2) / alpha. Here mu, sigma, and alpha are from normexp.fit.control(samplel_ElistRaw, ....) ----This is not the last step of the nec function but the later formulas after this, including the addition of the exponential term, does not change the first serval digits. Therefore, if my raw protein signal and mu are both integer and (sigma^2) / alpha is the same decimal number for one sample, then I shall expect to see these consistent digits after decimal after the correction.

But just given that the numbers look unusual to me, I wonder if anyone else experienced this and how you ended up explaining this consistency in your paper. It will be extremely helpful for me! Thank you so much!

Any thoughts and discussions are welcomed! I can also post the full versions of the formulas used in nec that I extracted from the Limma package if that will be helpful.

Sincerely, Ai

nec computes one set of negative control parameters (mu, sigma, alpha) for each sample. For values much larger than the background, the effect is to subtract a constant value, so getting the same decimal places is to be expected. For values closer to the background, the correction will be graduated. No one will care about the decimal places in your data or expect to you to explain them in a paper.