Also, I know that the "minus_germline" sample of the firebrowse data is created by comparing the tumor data to a standard blood derived normal sample that is estimated. Is it necessary to compare each sample to its original blood derived normal sample that is unique to each patient , on top of the standard comparison this website has already done?
My goal is to be sure that there is no chance of any private natural copynumber changes from the patients data so we know what is really a tumor related change.
You are evidently referring to the copy number data that has been processed by the Broad Institute and put on their FireHose / FireBrowse server. That data is taken from the Genomic Data Commons (GDC) and represents segmented copy number data. The segmentation algorithm is Circular Binary Segmentation (CBS) and the tool used in DNAcopy, as far as I am aware.
It makes sense that it's different from that which is hosted on the GDC as there is no synchronisation between both sites (the GDC updates the data over time), and Broad Institute may have applied additional filtering.