 |
 |
As luck would have it, I have put together a "challenge" data set series here: http://smb.slac.stanford.edu/~holton/challenge/twin/ Currently, the top contender is SHELXD, which can find all 12 sites using data with a twin fraction of 0.44. Phasing, however, still fails at this twin fraction. Nobody has been able to solve this structure using data with alpha=0.30 or higher. But, if you "cheat" and know the right answer, then all 12 sites are clearly resolved up to a 50:50 twin. So, theoretically, there is no reason why you couldn't write a program to do this, but to the best of my knowledge nobody has. Any takers? -James Holton MAD Scientist On 4/3/2016 2:05 AM, Eleanor Dodson wrote: > hmmm - in the ideal world with perfect data and a twin factor < 0.5 > you would detwin your data before trying sad or mad.. > > But detaining inevitably increases the Sigmas so a weak signal can be > lost.. > The lab philosophy was - if you need to do expel phasing, and the data > are twinned look for another crystal … But that may be too negative. > > You could try it and see if the phasing improved. > > Eleanor > > On 1 April 2016 at 21:40, Keller, Jacob <[log in to unmask] > <mailto:[log in to unmask]>> wrote: > > Dear Crystallographers, > > It occurred to me that SAD/MAD would work much better by > incorporating twinning into the HA finding/refining/phasing stage, > since the Bijvoet differences could be split according to the > twinning fraction and/or the HA's would be transformed by the twin > operator and weighted by alpha and (1-alpha). I believe twinning > is not currently modeled during this stage, even though for > twinned structures this would dramatically improve things almost > to the level of untwinned data, I think. Twin fractions could be > determined at the HA stage, then everything would go more > smoothly. Any thoughts of such an implementation? Perhaps Shelx > does this since 1969 or something? > > I thought of this because I saw some really deep "holes" in the > Phaser output from a twinned dataset, and think these might be the > evil other twin sneaking in? > > JPK > > > ******************************************* > Jacob Pearson Keller, PhD > Looger Lab/HHMI Janelia Research Campus > 19700 Helix Dr, Ashburn, VA 20147 > email: [log in to unmask] <mailto:[log in to unmask]> > ******************************************* > >