I have a single receptor and I have docked many compounds to it and I want to perform funnel metadynamics to the best three compounds. my question is how to determine the different parameters of funnel metadynamics for each one of the three complexes?

Should I perform a non biased MD simulation for each one then measure the standard deviation of the collective variables I am interested in?

or should I do an MD simulation for the cocrystalized compound (which is different from the top three compounds) then measure the standard deviation of the collective variables of interest and use it with the three complexes?