These are excellent questions. There are probably no easy answers. The situation with flexible loops is that the peptide chain adapts multiple (>=3) conformations here and we traditional crystallographers are trying to fit one model into the density that is the result of multiple confirmations. This is obviously a wrong thing to do.
For me, if there are densities for most main chains, I'd go ahead and built the whole loop (because I hate to have too many gaps in a structure). Here a good geometry is much more important than fitting all the densities since a single model cannot account for all the densities. A loop with a good geometry (sometimes you can get this from a library in various software packages) likely represents one of the multiple conformations. So at least it is true some of the times. Happy model building! Hong -----Original Message----- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of RHYS GRINTER Sent: Wednesday, October 17, 2012 11:06 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] Modelling Flexible Regions Hi All, This is a fairly basic question coming from a novice crystallographer, but I was wonder at what point it's reasonable to say 'I can't model that loop!'. I'm currently working on a couple of structures and there's density for where the a flexible loop on both structures goes, but there isn't good density for side chains (even aromatic ones) and even if the loop can be closed it doesn't seem like the Ramachandran plot will ever be happy. Both data sets are around 2A so it's not a question of poor density overall I was wondering if there was a consensus for when density is unmodelable, or if there are any tips for fitting residues into poor density. Thanks in advance, Rhys Grinter PhD Candidate University of Glasgow
