How to use Sobol analyses to apportion uncertainty in multiple “non-influential” parameters

Question

Having screened 54 parameters (Morris method) and identified 5 influential parameters through further analysis (Sobol), how can I then perform a Sobol analysis on the 2 groups (