Hi Huiying, Thanks, I think I understand what your approach was. I guess I was overly-concerned about model bias during density modification of model phases (I can explain to myself both why it's valid and why it's not; but haven't been able to resolve the obvious contradiction there); although with two mad phase sets in the mix as well it probably wouldn't have a large effect in any event.
I'd also thought that sigma_a (and other weighting schemes) reduced model bias, but that it wasn't possible to completely remove it. It's been a while since I read the sigma_a paper, thought. But seeing as your map improved, then these apparently aren't significant problems. Thanks again, Pete > Hi Pete, > > The sigmaa-weighing scheme implemented in SigmaA routine is the very means > to remove the potential model bias. Also, the model phases we used in the > phase combination were simply from a backbone poly-Ala model generated > from the best parts of the MAD-phased density (some of them from an > ARP/wARP run) leaving out the surface regions where the map quality is > marginal. Combining partial model phases with the experimental phases has > been used as a way to improve the map quality while the partial model is > still far from complete and most of the side chains are not yet filled > in. Once enough scattering mass has been built into the model and a > reciprocal space refinement with CNS or REFMAC is warranted, the > resulting 2Fo-Fc and Fo-Fc maps are often having much suprior quality for > the further model building. > > One drawback of combining partial model phases with the density-modified > experimental phases is that one cannot run the density modification second > time after the combination. I have not tested whether combining model > phases with the density modified MAD phases produces good quality map > (experts in the field can make comments). I did get significant > improvement in the map quality with the combine-then-modify procedure. > > HTH, > > Huiying > > On Fri, 20 Apr 2007, Peter Adrian Meyer wrote: > >>> structure" running mode of SIGMAA. This is the run we really wanted to >> combine >>> the model phases with the MAD phases before going through further >> density >>> modifications with SOLOMON or DM. >> >> I would have thought that you'd want to do this the other way around >> (density modification on MAD before model phase combination) in order to >> reduce possible model bias. >> >> I'm curious...what's the reasoning for doing the model phase combination >> first? >> >> Pete >> >> >> Pete Meyer >> Fu Lab >> BMCB grad student >> Cornell University >> >> >> >> > > -- > ------------------------------------------------- > Huiying Li, Ph. D > Department of Molecular Biology and Biochemistry > Natural Sciences I, Rm 2443 > University of California at Irvine > Irvine, CA 92697, USA > Tel: 949-824-4322(or -1953); Fax: 949-824-3280 > email: [EMAIL PROTECTED] > -------------------------------------------------- > Pete Meyer Fu Lab BMCB grad student Cornell University
