I would back way up in the process and verify that the basics are correct. Ten molecules in the ASU is unusual unless your molecules are forming defined oligomers, e.g., 2.5 tetramers, etc. Precise predictions of monomers in the ASU by Matthews number is increasingly unreliable above 2-3 monomers. You have to look at lattice packing to be more sure of the correct number. If there are defined oligomers, e.g.dimers, the initial search may work better with oligomers as the search unit than monomers.
How confident are you of the space group and unit cell assignment? What does the lattice packing of the MR solution look like? (You can do this in Coot.) Do protein molecules in the lattice make sensible contacts? Are there serious overlaps? Can you see solvent channels? Can you see symmetry-related oligomers? I'd sort this out before proceeding further. If these check out, then there are ways of using noncrystallographic symmetry in your ASU to significantly improve your initial maps for rebuilding, although sequence assignment may still be challenging at this resolution. Roger Rowlett On Mon, Aug 19, 2019, 8:15 PM Zhu Qiao <jasonqia...@gmail.com> wrote: > Sorry for the initial message. I tried to attach Matthews coefficient > calculation figure but failed to do so, which resulted the message as not > plain text. Below is my question, thanks. > > I collected one dataset and processed it to 3.6 angstrom. my protein is > quite small with only 14 kDa. It is estimated over ten molecules in one ASU > based on Matthew coefficient calculation. > > However, only ten molecules can be correctly placed with good fitting. I > can observe extra Fo-Fc electron density maps there needed to be modelled. > But Fo-Fc electron density maps are discontinuous. > > I tried to fix the ten molecules as the partial solution in phaser and > search more molecules, either resulted in no solution or the newly added > molecules didn't fit in the map after refinement. > > So I manually built the poly alanine chain in order to decrease the R > factor. I built around 200 amino acids into the final model. but these poly > alanine model can hardly be interpreted to remodel as my target protein > because of the low resolution. Currently the ten molecules model has a > Rwork/free as 0.33/0.36. The model with poly alanine chain has a > Rwork/free as 0.30/0.34. I can already extract the useful information > based on the well fitted ten molecules. And the fitting of the poly alanine > models could just for better model refinement. > > Can I get some suggestions regarding this kind of issue, what's the > general practise for such situation. Can I deposit the model with poly > alanine fitted and labelled it unidentified to pdb? > > Thanks for any suggestions and replies. > > Best Regards > Qiao Zhu > > ######################################################################## > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 > ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
