What is the best way to calculate

I would say that the calculation part follows the finding candidate moves part. - With no candidate move, there can hardly be anything to calculate, is it not? 

The first step is to find canidate moves.  This is normally done by pawn structure, moves played in similar positions, and opponents threats.  You should only find a handful of good moves.  Mentally play one of the moves and find the canidate moves for the side on move.  Repeat the process until the position is clear enough to provide a good evaluation.

With a calculator, OBVIOUSLY.

Always keep an abacus with you.

Calculation is moving the pieces. Visualization is seeing the future position clearly. Evaluation is judging the current and future position(s). Put it all together and that's analysis.

Very briefly speaking, candidate moves should satisfy both tactical and strategic considerations. I.e. they're safe against opponent's forcing moves, and they're in keeping with the needs of the position.

How to do all this is pretty fluid during real games in spite of authors like Kotov and Silman trying to give students to-do list in their books. In no real order you do a little evaluation, you do some shallow calculation, and some imagination too... after that you'll have some candidate moves to calculate. When you have multiple end positions with an eval, choose the candidate move that led to the better eval.

the best way to calculate is by looking at the most practical moves in each position and assessing the immediate outcomes of the next few moves.