Comment on "Using quaternions to calculate RMSD" [J. Comp. Chem. 25, 1849 (2004)]

Coutsias et al. have recently published a method to find the optimal rotational superposition of two molecular structures, which is based on a representation of rotations by quaternions (J. Comp. Chem. 25(15), 1849 (2004)). The method, which has been suggested by other authors before, is compared to the one by Kabsch, where the elements of the rotation matrix are directly used as variables of the optimization problem. The statement that the two methods are equivalent is misleading in the sense that the Kabsch method may yield an improper optimal rotation, which must be explicitly checked for, whereas the quaternion method does not mix proper and improper rotations. Nevertheless, both types of solutions can be considered by solving the same eigenvector problem. The relation between the two types of solutions is briefly discussed and bounds for the eigenvalues are given.