Hi Nick,

I looked through the files you sent and it looks like a bug in phenix.refine (someone reported the same problem before). Sorry about this. I fixed this bug yesterday so please use the latest PHENIX version from nightly builds available after January 17:

http://www.phenix-online.org/download/nightly_builds.cgi

Meanwhile a quick work-around is to remove a PDB file header (all REMARK 3 records) from input PDB file.

Also, to make sure using twinning in refinement will make any difference (at least R-factor wise), run this command:

phenix.model_vs_data model.pdb data.mtz

and see it it tells the data is twinned.

Please let me know if you have any questions.

Pavel.

On 1/17/11 11:53 AM, Nicholas Cody Sennett wrote:

Can phenix detwin data? I have tried to input the twinning law (l,-k,h) but phenix.refine does not create the appropriate maps and there is no appreciable change in R-factor. I have attached my xtriage and refinement logs.