Um... why are you doing this? Unless you want to try and increase consistency and yield on large scale, or somehow replicate a cheese exactly (which is super hard even if you do everything right), trying to do so much milk manipulation does not necessarily equate to the best quality cheese. Dry milk, for example, to me always tastes bad in cheese. I can always taste a defect. I would rather change up the cheese style or not stick to the identity so much, and focus on moisture and acidity control, and make cheese with the best, honest milk you have.
If you really want to know... yes casein:fat ratio is ideal measure because it helps to determine coagulant strength, moisture level, and final cheese FDM as a result. There's a table I posted here before, and one on UGuelph site. Yes, it's about 80-ish percent casein in the total protein fraction of milk. for NFDM addition, it depends on your source and what you get. The 2% comes from a guideline to avoid flavor defects. Yes, it's by weight, because you are calculating the total anhydrous casein.