On Sat, Sep 3, 2011 at 1:25 PM, Francis E Reyes wrote:

When specifying exptl_fobs_phases_freeR_flags_1.mtz (HLAM,HLBM.HLCM,HLDM) as an experimental phase file in phenix.refine via the .def file,  these HL coefficients are written to *_refine_data.mtz as HLA,HLB,HLC,HLD.

Now forgive me if I'm incorrect but HLA,HLB,HLC,HLD != HLAM,HLBM,HLCM,HLDM ? (where the former should come from phaser_1.mtz and the latter from resolve_1.mtz reflecting the density modification step).

Phenix doesn't really track column labels internally, or propagate them - it sees Hendrickson-Lattman coefficients, and automatically labels them HL*. The method that handles this isn't even capable of handling "HLAM" without modifying the code - the "A" will always end up at the end. Actually, Phenix doesn't organize data in "columns" either, so it's pretty detached from MTZ format. I'm not sure what the distinction between HLA and HLAM is - your interpretation sounds correct, but that would mean that AutoSol is writing out the density-modified phases instead of the experimental phases. Which doesn't sound right to me; I guess you could use phaser_1.mtz to be safe. -Nat

Hi Francis, Also it's not desirable to create a _refine_data.mtz as it's unclear where

the experimental phase labels came from the first place. Suppose I'm doing iterative rounds of MRSAD (with updated models during model building), I may have several AutoSol directories containing MRSAD runs for updated models. How am I going to know where the HL's in *_refine_data.mtz came from? (grepping .def files?)

The full file names should be in the .eff file. It may be useful to somehow store them in the mtz file, too. I'll keep it in mind.

It would be easier if the .def's contained the full paths of the original .mtz files and not a phenix combined data file. I may have my original reflections in one mtz, an R-free set in another, experimental phases in another, etc.

That's exactly why I decided to consolidate all inputs in one central file. In addition, you need some place to store newly generated R-free flags. Ralf