MHC-I peptide binding affinity is necessary in IP-MS?

Hello,

I get started to analyze MHC-I peptides from IP-MS data, recently.
In the research field, they usually use NetMHCpan to predict MHC-I binding affinity even IP-MS (if you are not familiar with MS, it is okay to focus on IP using MHC-I antibody)

So here is my question: why they predict MHC-I binding affinity in IP-MS? By definition, IP implies selecting MHC-I peptides for the purpose. Therefore, we do not need to predict MHC-I binding affinity because we already aggregated MHC-I peptides…

Here is my opinion: One of reasonable thought is that MHC-I antibody can bind non-MHC-I peptides (off-targeting) so we need to filter-out the off-targets. However, I cannot find the article about this issue.

Is anyone can comment the question?

I’m really appreciated to your effort in advance!