Maybe I need to elaborate on Ian's answer. SHELXD or other heavy atom location programs based on Patterson and direct methods have a 50% chance of getting either heavy atom enantiomorph, and in addition different solutions may be related by allowed shifts of the cell origin (8 in P212121).
This means that when you have found the heavy atoms, you must try both the original heavy atom coordinates and the inverted heavy atom coordinates for density modification. This might also involve inverting the symmetry operators (e.g. P41212 becomes P43212, but that is not a problem for P212121. For SAD only one heavy atom enantiomorph will lead to a sensible map. There is a special case in SAD phasing when the heavy atom arrangement is itself centrosymmetric, then both heavy atom enantiomers should lead to the correct map (e.g. one heavy atom in P21). SIR is less convenient. You still have to try both heavy atom enantiomers, but pure density modification will appear to work (or not) for both. However one of the maps will be a mirror image of the true map, e.g. the alpha helices will have the wrong hand. Automatic tracing programs like the SHELXE beta test will only trace the correct map, not the inverted map, so this is a quick way to find out which heavy atom enantiomer is correct. It is also the best check to see whether you have really solved the structure. However if the SIR heavy atom substructure happens to be centrosymmetric, density modification will not be able to break the enantiomer ambiguity. MAD and SIRAS will in general behave like SAD. However if your isomorphous difference is large and the anomalous signal is lost in the noise, they might be dominated by it and so tend to behave more like SIR. George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582 On Fri, 20 Aug 2010, Ian Tickle wrote: > Hi Pu > > Obviously your SAD & SIRAS solutions can't both be right, one must > have the inverted handedness: I would guess it's the SIRAS solution > that's wrong, since the SAD solution seems to have given you an > interpretable map. The reason for getting the wrong hand in the SIRAS > case is probably that the anomalous contribution is smaller > relatively, so is unable to discriminate the solutions. But what > makes you think that the SIRAS solution you (or the program) have > chosen is correct, in particular how do the peak heights in the > HA-phased map compare for the SIRAS solution inverted through the > origin, i.e. the SAD solution? > > -- Ian > > On Fri, Aug 20, 2010 at 5:35 AM, Pu Gao <ga...@moon.ibp.ac.cn> wrote: > > Hi all, > > > > I recently solved a structure using SAD or SIRAS successfully (refinement > > using native data). But I came across some questions about the heave atom > > (Hg) sites. SG : P212121. Cell: 61.000 137.700 142.170 90.00 90.00 > > 90.00 (native and derivative are very similar). > > > > 1. I used shelxD finding the Hg sites, but got different results from SAD > > and SIRAS. Here paste the first 3 sites: > > X Y Z X > > Y Z > > (fractional) > > (orthogonal) > > SAD: > > Hg1 0.598389 0.567345 0.098041 36.759 78.146 13.983 > > Hg2 0.901955 0.570168 -0.098881 55.407 78.535 -14.102 > > Hg3 0.727585 0.594368 -0.240330 44.696 81.868 -34.276 > > > > SIRAS: > > Hg1 0.598717 0.432159 0.098363 36.522 59.508 13.984 > > Hg2 0.901527 0.430199 -0.098437 54.993 59.238 -13.995 > > Hg3 0.727432 0.406845 -0.240253 44.373 56.023 -34.157 > > > > It seems there is a symmetry plane m(x,1/2,z) between SAD sites and SIRAS > > sites. The above HA coordinates were in real space not the partterson > > space, right? So could someone tell me why there is a m? > > PS: the Harker section plots (FFT for patterson in CCP4) of difference > > patterson and ano-difference patterson are very similar. > > > > 2. I put the Hg sites caculated by shelxD (say, the SAD sites) to SHARP and > > got the density map (looks very good). RESOLVE was used to autobuild the > > model. The Fo-Fc map (Fo from Hg-data, Fc from the model after refinement) > > at sig>5 obviously showed the Hg sites (around Cys). But the heave atom > > coordinates (from shelxD or sharp) didn't match the Fo-Fc density. I don't > > know why is that? Dose it bucause the coordinate system mismatch? > > > > Any help would be much appreciated. > > > > Thanks, > > > > Pu > > > >