Probabilistic and causal modelling of genome-scale data