Error when using flex-em
Dear Modeller support,
When trying to fit the crystal structure in an EM map using mod9v7 and the flex-em script (unmodified apart from filenames) I get the following error message:
bash $> /usr/bin/python ../../scripts/flex-em.py > flex.log File "/fbs/emdisk3/emsoftware/LINUX/modeller9v7/modlib/modeller/optimizers/molecular_dynamics.py", line 50, in optimize ret = func(self._modpt, mdl.modpt, edat.modpt, libs.modpt, inds, **vars) OverflowError: check_inf__E> Atom 1 has out-of-range coordinates (usually infinity). The objective function can thus not be calculated.
It happens when running in CG mode and MD mode. The map is 140x140x140 pixels, which is already 1/8 of the complete map (using the complete map gives the exact same error). There are 893 residues (6 rigid body domains) in the model I am trying to fit. It gives the same error with poly-Ala. Are these files too large? The coordinates of the pdb files that are output after the first iteration look sensible and fit well in the map (but not quite well enough). I would be very grateful for any suggestions to help solve this problem.
Many thanks,
Saskia
On 2/23/10 5:06 AM, Saskia Bakker wrote: > OverflowError: check_inf__E> Atom 1 has out-of-range coordinates > (usually infinity). > The objective function can thus not be calculated. > > It happens when running in CG mode and MD mode. The map is 140x140x140 > pixels, which is already 1/8 of the complete map (using the complete > map gives the exact same error). There are 893 residues (6 rigid body > domains) in the model I am trying to fit. It gives the same error with > poly-Ala. Are these files too large?
This means that the forces on the system were really large at some point, which forced the coordinates to shoot off to infinity. This is usually an indication of a problem in your scoring function - perhaps two atoms were placed at the same position, for example. It's difficult to suggest a solution in your case, since it sounds like your system is OK, but you can certainly try modifying the rigid body regions or optimization parameters.
Ben Webb, Modeller Caretaker
participants (2)
-
Modeller Caretaker -
Saskia Bakker