I'm trying to use the *IMP.pmi.tools.set_coordinates_from_rmf *to start
modelling a complex from a previous modelled frame.
This will only work if the topologies of the two complexes are the same.
In a first step, I
set the whole structure as a single rigid body and perform 100 rex
frames. In the second step I want to use the frame 99 of this modelling
as starting point, but now I want to divide the structure into several
rigid parts and super rigid bodies.
This won't work because the RMF file doesn't save the global coordinates
of each particle in the rigid body - it saves the reference frame of the
body and the body-frame coordinates (relative to that frame) of each
particle. So there's only one reference frame in the RMF (for the one
rigid body in the original modeling) and you're trying to apply it to
multiple rigid bodies. If you have the same set of particles in both
cases, you should be able to read in their global coordinates, then map
them to body-frame coordinates for each new rigid body. (There are
methods in the RigidBody class to do this.) Or you could just set all
your new rigid bodies to have the same reference frame. But this would
be an "exercise for the reader" - there is no code in PMI (to my
knowledge) to do this.
Another option would be to write out PDB or mmCIF files for each subunit
from your first simulation, then use these in the topology file for the
second one.