Dear Ed, Many thanks for this careful explanation, which I appreciate. I realise in my own practise on such matters I have two situations :-
(i) where, albeit limited, electron density evidence, coupled with chemical evidence, leads to an attempted model atomic fit but the B factors there sky rocket. (ii) there is no electron density evidence and even though the chemical evidence is sound one can in fact add nothing. Here I simply concede that there is nothing one can do. The issue here is not whether to keep these atoms but simply you really cannot make a start finding them. Greetings, John On Fri, Apr 1, 2011 at 4:03 PM, Ed Pozharski <epozh...@umaryland.edu> wrote: > Dear John, > > there may be reasons to disagree with both options. This has been a > recurring discussion for many years, and in my mind the most convincing > arguments for both sides are as follows: > > "Keepers": > > I know the side chain is there and the high ADP is a good approximation > of reality. Removing atoms causes such a mess for the end user. > > "Deleters": > > We don't model missing loops, termini, ligands and waters when there is > no density, and side chains should not be treated differently. Most end > users think ADP is a nucleotide and will over-interpret the model. > > I am a "keeper" when it comes to end user treatment, but a recently > converted "deleter" when it comes to modeling (a rather stressful > position). So I am not taking sides really, but rather looking for a > middle way. (Have to admit that my secret goal was to knock down the > zero occupancy fallacy :) > > Perhaps these ideas are worth exploring: > > 1. Provide dual representation - a crystallographic model and an > end-user model, both downloadable from the PDB. > 2. Model missing side chains "NMR-way" > 3. A new data file format is needed (mmCIF?) that combines atomic model > with electron density, and visualization/analysis software shall be > modified to always utilize the experimental data > 4. Implement reduced ADP restraints for disordered side chains to > further reduce model bias > > But ultimately, as long as experimental data is deposited, I believe > that people are free to interpret their data the way they see fit. > Others are then free to look at the electron density and become outraged > at the interpretation. > > Cheers, > > Ed. > > On Thu, 2011-03-31 at 23:25 +0100, Jrh wrote: >> Dear Ed, >> Thankyou for this and apologies for late reply. >> If one has chemical evidence for the presence of residues but these >> residues are disordered I find the delete atoms option disagreeable. >> Such a static disorder situation should be described by a high atomic >> displacement parameter, in my view. (nb the use of ADP is better than >> B factor terminology). >> Yours sincerely, >> John >> Prof John R Helliwell DSc >> > > -- > "I'd jump in myself, if I weren't so good at whistling." > Julian, King of Lemurs > > -- Professor John R Helliwell DSc
