The very best advice is: get better data if at all possible. Things are much easier and more informative at 3A than at 4.2A and even better at 2.5A.
There are various tricks which might help improve diffraction.. But if you are stuck then: The partial MR phases are very useful for checking your derivatives - look at difference Fouriers and anomalous Fouriers to see if anything has bound.. Then if you do have substitution Phaser SAD + MR will help to improve phases. If there is NCS then you can use Parrot or DM to improve the phases substantially. Buccaneer is quite good at building missing parts - we are experimenting with different scenarios to optimise this.. Eleanor On 14 February 2014 09:41, Guenter Fritz <guenter.fr...@uni-konstanz.de> wrote: > Dear Sun, > > I had a similar problem. If you have a good TaBr dataset this should give > you good phases. > You can combine the TaBr phases and Molrepl phases in SHARP. > What worked in my hands very well and is easy to do: SAD using Phaser and > the partial Mol Repl Model. > > You find here a input file for such a scenario: > http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.php/Phenix#phenix.phaser_-_SAD_phasing_with_Phaser > > Or simply use the CCP4 gui Phaser and 'SAD with molecular replacement > partial structure' > After DM or parrot you might get some interpretable density even at 6.5 A. > > If you don't have a good Mol Repl Model for the missing part you might try > http://toolkit.tuebingen.mpg.de/hhpred > to look for structures that are not that closely related. > > SeMet can help a lot in this case too. You will get the SeMet positions at > low resolution with Phaser as described above. > The SeMet positions will guide you to place a model into a very crude > density. > > HTH, > Guenter > > > Dear Crystallographers, > > I'm working on a ~90KDa membrane protein, with big extracellular part, > probably function as dimer. > Now we have dataset to ~4.2 Angstrom and using extracellular homolog > structure we can find a solution for this part(~45% of the whole molecule > MW) through molecular replacement, and the molecules are packed as layers, > and the other part are presumably between these layers. However, we are > having trouble to fit the rest of the protein even though there're some > density between the solved part. Rfree is at 40% now. > > We're trying to do heavy atom soaking, such as TaBr. We collected data for > MIR but it's not helping so far. (Can I combine these MIR data with the > native dataset because the MIR set is only at ~6.5 Angstrom)? > > Other information: this protein is expressed in Sf9 cells (so very hard to > do Se-Met derivatives). The crystals is nice and big and cubic. > > Any suggestions or examples? Thanks a lot. > > Bingfa > >
