Dear James,

> Phasing, however, still fails at this twin fraction.  Nobody has been
> able to solve this structure using data with alpha=0.30 or higher.

Did you try the latest version (2016/1) of SHELXE?

I don't know how you define "solve this structure", but
it was trivial to get an interpretable map at least up to
alpha=0.40 using SHELXD, SHELXE and Buccaneer. I uploaded
the results to my Google Drive.
https://drive.google.com/file/d/0ByqmCYlnqv6UeHBxNS1oalRXeUk/view?usp=sharing
As you see, auto-traced models became less and less complete
with increasing alpha, but the map is still clear enough for
experienced crystallographers to fix the model.

Below is how I got these maps and models.

First, I converted mtz files into sca files by
phenix.reflection_file_converter.

Then, finding 12 sites was easy with SHELXD using the following
input.

-----------------------------------------------------------------------
CELL 0.9340 52.658 52.658 217.059 90.00 90.00 120.00
SPAG P65
FIND 12
SFAC SE
NTRY 20000
SAD inp.sca
-----------------------------------------------------------------------

I put NTRY 20000 but killed SHELXD earlier when the best CC stopped
improving.

Next, I run SHELXE with many combinations of resolution limits
and solvent contents as I did for SFX S-SAD phasing (Acta D, 2015).
I used the following script.

-----------------------------------------------------------------------
for reso in 2.0 2.1 2.2 2.3 2.4; do
   for solv in 0.50 0.47 0.45 0.52 0.55 0.57 0.60; do
      for z in 12; do
         ln -s sad.hkl reso$reso-solv$solv-z$z.hkl
         ln -s sad_fa.hkl reso$reso-solv$solv-z${z}_fa.hkl
         ln -s sad_fa.res reso$reso-solv$solv-z${z}_fa.res
         qsub -cwd <<EOF
shelxe reso$reso-solv$solv-z$z reso$reso-solv$solv-z${z}_fa -s$solv -a25 -h$z -z$z -d$reso -q
EOF
      done
   done
done
-----------------------------------------------------------------------

I checked the results by "grep CC *.pdb | sort -k7" and fed
promising results into Buccaneer using the following script.

-----------------------------------------------------------------------
SEQ="../3dko.fa"
CELL="52.64,52.64,216.99,90.00,90.00,120.00"
SPAG="170" # P65
if [ -z "$RESO" ]; then
   RESO="2.2"
fi

convert2mtz -hklin $1.phs -mtzout $1.mtz -cell $CELL -spacegroup $SPAG -colin "FO FOM PHI SIG"
mkdir $1-buccaneer

buccaneer_pipeline -stdin > $1-buccaneer.log << EOF
pdbin-ref $CCP4/lib/data/reference_structures/reference-1tqw.pdb
mtzin-ref $CCP4/lib/data/reference_structures/reference-1tqw.mtz
colin-ref-fo FP.F_sigF.F,FP.F_sigF.sigF
colin-ref-hl FC.ABCD.A,FC.ABCD.B,FC.ABCD.C,FC.ABCD.D

seqin $SEQ
mtzin $1.mtz
colin-fo FO,SIG
colin-free FreeF_flag
colin-phifom PHI,FOM
pdbin $1.pdb
pdbout $1-buccaneer.pdb

cycles 5
buccaneer-anisotropy-correction
buccaneer-fix-position
buccaneer-fast
buccaneer-1st-cycles 3
buccaneer-1st-sequence-reliability 0.95
buccaneer-nth-cycles 2
buccaneer-nth-sequence-reliability 0.95
buccaneer-nth-correlation-mode
buccaneer-new-residue-name UNK
buccaneer-resolution $RESO
buccaneer-keyword model-filter-sigma 3.0
buccaneer-keyword mr-model-filter-sigma 3.0
refmac-mlhl 1
jobs 4
refmac-twin 0
refmac-keyword REFI RESO 100 $RESO
prefix $1-buccaneer/
EOF
-----------------------------------------------------------------------

Best regards,

Takanori Nakane