Hi list,

In order to have a better insight I modified my script, keeping only the em.FitRestraint in my model (script attached).

The score for the initial model (5.96046447754e-08) is reassuringly near 0; which is reasonable when comparing two maps sampled from the same structure.

Nevertheless, when chains are randomly moved, the score does not radically improve, even when the transformations move them far away from their initial position. Moreover, I have an example where the score is minimized from 0.37 for the initial (moved chains) position to 0.49 for the optimized solution !?

I may have missed something, and I would need some technical support to fix the problem quickly.

--Ben