Often in cáncer related studies people use cut offs for cnvs (e.g. >=2x fold change for duplications and <=0.5 fold change for deletion). It is clear why it is done for deletions, the variant in tumor suppressor has to be homozygous, but why do they use this threshold for amplifications? Expression changes significantly even with copy number 3 instead of 2...
The main version so far is because of level of noise (CN3 can be less reliably detected than CN4), but when we look at sub-clones, it becomes less obvious. Also, modern tools provide methods for False positives assessment.
So, the question is: why do everybody does hard thresholding for amplifications, is there a hidden biological sense or it is just because bioinformatics methods and data were not reliable back in times?