Hi Daniele, I'm not sure that there is a universal answer to this. I would just try and see if the result makes sense. In phenix.refine you can do it by adding "group_anomalous" to "strategy" and specifying what are the anomalous scatterers to be refined (use "anomalous_scatterers" scope of parameters for this). See "f' and f'' refinement" section of phenix.refine documentation for details: http://phenix-online.org/documentation/refinement.htm Please let me know if you have any questions. Cheers, Pavel. On 12/2/2008 10:23 AM, Daniele de Sanctis wrote:

Hi,

I'm making some refinement tests with anomalous data and I wanted to try to refine the anomalous contribution. When would it be the best stage when to introduce it?

Thanks

D