Hello all, I am running a scheme script to do real-space refinement for GroEL, which is composed of 14 ~500-residue chains, i.e. it is quite large. I have noticed that the speed of Coot v0.3.1 has decreased significantly over the course of the run while the memory usage has grown ~3x. Am I doing something awfully wrong, or is Coot leaking memory? I've attached my scheme script and a revised version of fitting.scm (wrote a derivative of stepped-refine-protein called stepped-refine-chain). I'm running v0.3.1 on a Xeon cluster with 32-bit Ubuntu linux v2.6.21 and Mosix2 v2.21.0.0. Thanks, Nasos Dousis