Im having doubts on myself of just adding and removing some peaks for my research regarding the bio marker detection. (VRE versus VSE)

From the clinprotool software on Maldi-Tof, There's artefacts that can minimise through the data processing steps such as 4 algorithm and preprocessing steps like setting the spectrum, peak calculation, recalibration, average peak list calculation etc... about the manual peak picking ? What should we consider? Features  that based on relevance criterion, redundant features exist and strong discriminating factors....

And what does the model calculation? Do we need to set it by ourselves? From the literature I read it was set us default. And it was still distinguishable 

Please I beg for help. Thank you so much.