Sent on behalf of Kazuhiro Yamada ([log in to unmask] ) I am trying to use CNS ver. 1.2 for structural refinement of our crystal containing three 1:1 hetero complexes with pseudo-3fold symmetry in the A.U.([A1B1], [A2B2] and [A3B3]). Following initial model building and refinement the rotation matrices & translation vectors of the A molecules (from 1 to 2, or 1 to 3) are slightly different from the corresponding transformations for the B molecules due to a relatively loose interaction between the A and B components. Thus at the simulated annealing step I want to define two versions of ncs_matrix_2 (A1 to A2 versus B1 to B2) and two of ncs_matix_3 (A1 to A3 versus B1 to B3). However, there is no template describing such a case on the CNS ver1.2 homepage. Does someone who is well acquainted with NCS know how to address such a case? What should I do practically..., average the two sets of ncs_matrix_2 & _3? Or write a special subroutine / script to handle it? Thanks for any tips. Kazuhiro _________________________________________________________________________________________________________________________________________________ Dr. D.F. Sargent Institute of Molecular Biology and Biophysics ETH-Hönggerberg CH-8093 Zürich Switzerland tel (+41) (0)44 633 2487