To reply to the original question, "can the calculated Matthews coefficient be wrong?": have you read the paper by Matthews? There is a distribution of solvent content in macromolecular (protein) crystals. It has a "peak" and tails on both sides. Meaning that there are "outliers" in the distribution, i.e. crystals with very little solvent, and crystals with plenty of solvent. They are not really outliers since, as in any physical phenomenon, some data points fall outside 1 sigma or 2 sigmas from the mean. Exceptions do exist.
What you cannot have however, is an asymmetric unit that contains too many molecules/subunits than what is physically possible. Molecular replacement calculations can be performed searching for 1, 2, 3,... N copies in the asymmetric unit. This is what I personally do because at first I do not know the content of the asymmetric unit in my crystal. Fred. > Message du 25/06/10 07:54 > De : "xinghua qin" > A : [email protected] > Copie à : > Objet : Re: [ccp4bb] Matthews coefficient > > Hi everyone: > Thanks for all the responses. > The Matthaws-cell content analysis programe in CCP4 package gives the > results: 47% solvent content and 3 molecules in asu with 87% confidence. > the space group is P3121. how to carry out self rotation function? can > phaser do that work? If there areTwo moleculars in asu ,no clashes and the > TF values are above 10.but if three, the clashes are too much and TF valur > is about five. > And how to calculate the solvent content? Is the calculated solvent content > with the Matthews-cell content analysis programe not always right? > Best regards > > Xinghua Qin > On Fri, Jun 25, 2010 at 10:42 AM, Vineet Gaur wrote: > > > Hi > > It would b good if u can mention the solvent content and space group along > > with the no. of molecules in asu/ > > > > u can carry out self rotation function to check the no of mol in asu > > best > > vineet > > On Fri, Jun 25, 2010 at 6:42 AM, xinghua qin wrote: > > > >> hi CCPeers > >> The Matthews coefficient of my protein is 3 calculated with > >> matthews-cell content analysis CCP4 programe with 87% confidence , but when > >> doing the refinement the third molecular couldn't get into the unit cell > >> because of too many clashes.Deletion of the clashed AA did not work well, > >> Then I used two molecules in the unit cell, After refining with Refmac, I > >> found that the R factor is 0.29, R free is 0.40.I believe the value can be > >> better with the real space refinement.But the question is that can the > >> calculated Matthews coefficient be wrong? > >> > >> Best regards > >> > >> Xinghua Qin
