I see the do-calculus as being causality as applied to a graph. It then gives you a bunch of correlations to test. The other example of "things applied to gra

I see the do-calculus as being causality as applied to a graph. It then gives you a bunch of correlations to test. The other example of "things applied to graphs" is graphical models aka Bayesian networks => Bayesian statistics.