Embedded distance geometry (substructure) coordinates require extensive regularization. The ``dgsa.inp" protocol below uses template fitting followed by simulated annealing to regularize the coordinates. The protocol is close in spirit to the one published by Nilges, Clore, and Gronenborn (1988b). However, many improvements have been added to the original protocol, and it appears that the present protocol is more efficient and robust. The example file is general and should work for all types of macromolecules; it has been tested for proteins and nucleic acids. No change is required when using this protocol for nucleic acids compared to proteins. The starting coordinates have to be defined for at least three atoms in each residue (see the definition of atom tags in Section 2.15).
Disulfide bonds and other covalent links are treated as ``real" bonds in this protocol, in contrast to template generation.
Important parameters that the user might want to change are the simulated annealing starting temperature, the duration of the high-temperature stage, and the duration of the cooling stage. The example file worked well for proteins and nucleic acids with up to about 2300 atoms. An increase in the duration of the high-temperature and cooling stage is required if the initial distance geometry (sub)structure is far from the correct solution. There is an abort condition that is triggered when the temperature during molecular dynamics rises uncontrollably. This condition requires manual intervention, which involves decreasing the time step of the molecular dynamics portion of the slow-cooling stage of the protocol.
The filenames of the family of regularized
coordinates are
``dgsa 1.pdb" through ``dgsa 10.pdb".
The coordinate files
contain useful information about the energies,
restraint-satisfaction, and violations.