A model-fitting procedure incorporating the double-null technique for use in protein structure refinements

The double-null technique in conjunction with the conjugate-gradient algorithm gives an efficient method of regularizing protein structures. The double-null technique constrains planar groups by introducing two artificial atoms which are geometrically the apices of tetrahedra from three selected atoms in the planar group. The model-fitting procedure is simply concerned with reconciling information on (x, y, z) atomic coordinates with information on bond length, bond angle and planarity constraints. This can be done without treatment of hydrogen-bond, van der Waals or electrostatic interactions and consequently the minimization function is nearly quadratic. The equations which have to be solved have many unknowns but few coefficients and an effective iterative method can be used.