The experimental data can be classified in two ways. Classes of restraints usually correspond to a single spectrum (mixing time). A group of classes corresponds to a group of spectra measured under the same physical conditions (relaxation network, solvent, magnetic field strength, temperature). This distinction is made since, in principle, a change in the physical conditions makes it necessary to recalculate the relaxation matrix, while a change in the mixing time does not. Order parameters, correlation times, occupancy values, and the relaxation network can thus be specified for each group of classes separately.
The intensity data are read into a class, and then the group that the class belongs to is specified. The syntax is similar to the NOE assign statement (see Section 18.1), but the input format is more flexible, to allow the user to input NOE buildup curves and error estimates. Before issuing an assign statement, one has to specify how many mixing times are to be read in at a time and what format for the error estimates the program should expect. The number of classification statements determines how many mixing times are included in each assign statement. The error_input statement determines the format of the error estimates. Ambiguous NOESY cross peaks can be entered using multiple atom selections. During the refinement, the sum of the corresponding calculated intensities is restrained to the sum of the observed ones.
The example intensity file below shows a number of different options. Note that the class and error statements remain valid until other class or error statements are entered.