Hi, I am using molecular replacement to build a model against a 2.2A resolution data set. However, the template model only has 23% sequence identity. First, I use chainsaw in CCP4i to prune all side chain to be poly-Ala except the identical residues. Then I use AutoMR and rigid body refinement, I got Rfree=50%. The continuous electron density is not bad showing second structures. However, the model has some out-of-registration problem, for example, two proline are in a alpha-helix. In other words, it has model bias. I want to use composite SA-omit to get a less model bias map, not a simple composite omit map or a SA_omit map ommiting a part of the region. Does someone know who to do this? Thanks! Qun Wan postdoc fellow Case Western Reserve University