[phenixbb] phenix and refmac5

Nathaniel Echols NEchols at lbl.gov
Tue Dec 15 03:25:19 PST 2009

On Dec 15, 2009, at 11:02 AM, anon wrote:
> recently i used phenix (AutoSol and AutoBuild) on a SAD dataset containing Iodine. I was wanting to do my refinement using Refmac instead of phenix.refine. I was wondering where would i be able to locate the correct .mtz file to input into Refmac5 for rounds of refinement? I was under the impression the .mtz file used for refinement was the one produced after scaling, but im not sure where phenix puts this.

Look in the AutoSol output directory - there should be a file named something like exptl_fobs_phases_freeR_flags_1.mtz (the number is variable).  This is probably what you want to use.

> Also I was wondering if anyone had any issues with phenix.xtriage carrying out its Matthews coefficient calculation? When using Phenix, it gave me a solution with 3 mols/AU, despite there being 4 mols/AU according to the Matthews coeffiecient calculation in CCP4. Looking back at the xtriage log in autosol it has n_residues= 200 (despite me not specifying anywhere the number of residues, my protein has 147 residues). Realising this I re-ran Autosol, inputting my .mtz file along with my sequence this time, and specifying the number of NCS copies as 4. Looking at the phenix.xtriage log this time results in it recognising the correct number of residues, however the matthews coefficient calculation is still incorrect, claiming 1 mol/Au. Any ideas what may be the problem?
> scaling {
>  input {
>    asu_contents {
>      n_residues = 588
>      n_bases = None
>      n_copies_per_asu = 4
>    }

It looks like AutoSol is pre-emptively multiplying the sequence length by the number of NCS copies, but still reporting the number of copies separately.  The 1 mol/Au is based on the total number of residues reported.  If you run Xtriage manually, it should give the correct answer; this is probably something that needs to be fixed in AutoSol.

Nathaniel Echols
Lawrence Berkeley Lab
NEchols at lbl.gov

More information about the phenixbb mailing list