Re: [ccp4bb] NOT Job Advertisement but model refinement
This has been told (written, in fact) on this bb many times. The purpose of refinement is not to bring the R-factor down, but to obtain a model that satisfies best all available data (including biochemical data). Anyway, as to what you can do: it's always possible to question the choice of TLS groups. I do not know how you select yours. What I personally do is to connect to the TLSMD server, lauch the computation, then have a look at the output curve. There is a steep descent at first, and (usually) a break in the curve. This break in the curve I use to tell me how many TLS groups to be used during refinement. One can always change the refinement program to see if an improved geometry together with more appropriate R and R-free values can be obtained (both at the same time). But you do not give much information on how you have built and refined your model so far... Also, sometimes automated model rebuilding programs give excellent results. Fred. atul kumar wrote: dear all I am trying to solve structure of data set at 1.9 A,r merge 9.3,it belongs to space group P3221 unit cell 160.78 157.32 135.62 90.0 90.0 90.0,i has no NCS ,i did tls as well but after water addition the r factor and r free is stuck at 22 and 25 respectively. kindly guide me how can i bring down the R factor. Thankin you. Regards Atul
Re: [ccp4bb] NOT Job Advertisement but model refinement
atul kumar wrote: dear all I am trying to solve structure of data set at 1.9 A,r merge 9.3,it belongs to space group P3221 unit cell 160.78 157.32 135.62 90.0 90.0 90.0,i has no NCS ,i did tls as well but after water addition the r factor and r free is stuck at 22 and 25 respectively. kindly guide me how can i bring down the R factor. Maybe they're stuck because the refinement is already as good as possible for these data? R-factors of 22/25 aren't unreasonable at 1.9A, so unless you see other problems with the structure (suspicious difference density blobs, missing residues, poor geometry, etc.), I'd stop worrying. Inspect the maps for missing pieces, feed the structure to the Molprobity server, fix as many problems as possible, do one last refinement run, and move on to the next experiment, or start writing a paper. -Nat