How about phase extension using DM, sure you say you only have one mol per asu but it might still be worth trying various approaches of solvent flattening/flipping. Don't know what you used to detect your sites and refine them, but it also might be worth sticking them into Sharp with your partial model and see if the phases improve. You say you have 450/750 residues what makes you believe that you placed them correctly ? Also do you have space from the crystal lattice packing for the additional 300 residues ? In other words are you certain that what you have crystalized is the full length version of your protein ? Adding side chains leading to worse Rfactors would suggest that you are most likely not in frame. Since you have a partial backbone you could try finding a "homolog" via EBI SSM to serve as a better starting model, aligning your sequence to it and replacing the side chains accordingly.
Another suggestion keep changing programs don't stick to Refmac visit Phenix or the other way round, and as a last resort you could give GraphENT a try. Jürgen On Sep 1, 2011, at 3:00 PM, Pete Meyer wrote: Hi, Depending on how many zn sites you have, you may be able to do zn-mad for your native crystals. You don't mention if you've tried combining your various sources of phase information; if not, it's worth looking into. You may also want to look into various multi-crystal techniques (averaging, phasing and/or merging) - I've had decent luck with multi-crystal phasing off zn at low resolutions. Good luck, Pete Basudeb Bhattacharyya wrote: Dear all, We're looking for some advice about how to proceed with a structure we're working on. Our protein is 750 amino acids and naturally binds zinc. We have a SeMet data set that goes down to 3.7 angstroms. 4 of 8 selenium sites are ordered and visible in addition to our zincs and we've modeled about 450 residues of C-alpha backbone off a pure SAD density map (we've tried other phasing experiments such as Zn SAD, MAD, etc. and the best maps--clear density and visible secondary structures--we get are off Se SAD). We have one monomer per AU (and we have secondary structure coverage over our entire protein based on looking at conserved domains of our protein--unfortunately, MR is not working for this project). R/Rfree hits a minimum of 37/41 respectively in spacegroup P32 (we've tried P3 and P31, which haven't worked). We also have a native set down to 2.7 angstroms. We are able to place our working model into the native data set, but we are unable to further refine the structure in Refmac (density doesn’t improve and the stats creep up). Addition of side chains only makes our stats worse. The data sets are clean (no twinning, etc.). While we understand that we may need more phasing information (i.e. our initial model may still be quite inaccurate given resolution and size of the protein among other things--we are trying to improve this), we're wondering if anyone might have some other suggestions or insights about how we can move forward given the data that we currently have. Thanks in advance for any advice. Sincerely, Basu ...................... Jürgen Bosch Johns Hopkins University Bloomberg School of Public Health Department of Biochemistry & Molecular Biology Johns Hopkins Malaria Research Institute 615 North Wolfe Street, W8708 Baltimore, MD 21205 Office: +1-410-614-4742 Lab: +1-410-614-4894 Fax: +1-410-955-2926 http://web.mac.com/bosch_lab/