Hi Oliver, I'm not sure what's going on, since from what you described it seems like you doing what one should do. If you send me the data and model I will be happy to have a look. The fact the you get substantially lower R-factors in Refmac tells me that there must be an obvious fix in phenix.refine or in what you are trying to do using phenix.refine so it produces comparable or better R-factors. Pavel. On 5/7/09 12:36 AM, Oliver Clarke wrote:

Hi,

I'm running phenix.refine v1.4-3, and I'm refining a 3.3A structure that has potassium ions present on a crystallographic fourfold axis. I believe (based on previous refinement in REFMAC5.5), that these ions are at full occupancy, so I set the occupancy of each ion to 1.

Despite this, there is still a lot (several 7-10sigma peaks in the Fo- Fc) of unexplained positive density, the potassium ions do not converge to their previously refined positions, and their B-factors refine upwards to values several times higher than the surrounding protein. When the same structure is refined in refmac (with ion occupancy set to 0.25), there is little to no unexplained density and the R/Rfree are substantially lower (21.9/25.9 as against 24.5/31)

Does anyone have any ideas/suggestions as to what might be going on?

Cheers, Oliver _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb