Hi Edward, in our case, the analysis performed with Bruker Dynamics Center does provide an overall rotational correlation time consistent with the mass of the homodimer in solution, and the proportions of the calculated ellipsoidal diffusion tensor are also consistent with the structure of the homodimer. Here are a couple of references in which the authors applied the model-free treatment to a homodimeric protein, apparently without worrying too much: Biochemistry 2000, 39, 9108-9118 Biochemistry 2003, 42, 1890-1899 Any feedback is most welcome. Regards, Stefano
For the homodimer problem, you will probably have to research that one. The problem here is that you are measuring different data each monomer but that this data is averaged via the NMR spectrum. If you apply the analysis to the full dimer structure, you are essentially saying that the non-averaged data for each monomer is identical to the averaged data. If the averaged data should not be used for the monomer, it would be just as bad using it on the dimer. From memory there are papers investigating this. And from even more distant memory, I think I remember someone invoking symmetry arguments - though I don't remember if this was in a paper, a discussion at a conference, etc, or even what the conclusions of those symmetry arguments were. I would be interested to hear if you know more about this problem and what the solution is. I'm sorry I cannot help with this relatively rare problem. The key however will be the symmetries of the ellipsoidal diffusion tensor. Regards, Edward On 10 February 2014 15:56, Stefano Luciano Ciurli <stefano.ciurli@xxxxxxxx> wrote:Hi Edward and thanks for your message. Yes, I am looking forward to mastering your software. I hope I do not bug you with too many questions, and that the learning curve will be steep. See below:Welcome to the relax mailing lists! I have just tried out your PDB file attached to the bug report. I launched the relax in GUI mode and performed the following steps: - Selected a model-free analysis from the new analysis wizard (this is the special dauvergne_protocol auto-analysis which is worth reading about to understand what is happening). - Clicked on the 'Spin editor' button. - Clicked on the 'Load spins' button in the spin editor window. - Selected the option 'From a new PDB file' in the spin loading wizard. - Loaded your PDB file. - Set the spin ID to '@N' and '@H' to load the 15N and 1H spins of the backbone. '@NE1' and '@HE1' spins should probably be loaded as well. - Click on 'Finish'.I think I did all the steps that you describe here below yesterday, except fot the only fact that I changed the string about the molecule name, which seems unlikely to be the cause of my failure. Now, and I did it several times, I do not have any problems in reading the spins for the PDB file. OK then!I can now see the 15N and 1H spins for two molecules. To perform a model-free analysis, I could simply delete one of the two molecules and continue.question arises: the protein is a homodimer in solution, and I do not think I should delete any spin nor molecule. Am I correct? If I do, I start from a wrong geometric model, no? Stefano