Characterising copy number alterations


I have been using the Cell Model Passports Copy number alteration “CNV Driver gene data” ( dataset.

I am a bit confused and would like clarification with how the CNV scores are calculated ( as some of the values are 2 or 3 but it has been characterised as a “loss” when the ranking system says 2 = neutral and 3 = gain.

Thanks in advance for any help