> Temperature control: > I did have a recent temperature calibration in place, but did not calibrate > when setting up the experiment. > I probably should have done that. > > I did use heat compensation blocks during the experiment, but did not do > single scan or experiment interleaving. > > Is there a way to account for this in relax, or is re-recording of the data > the only remedy?
As I mentioned to Thibault, book some more NMR time (~2 hours) and rerun mini-versions of your experiments with MeOH or methylene glycol standards. Then measure the temperature of the experiment. But if you did minimally use fid interleaving for all experiments (single scan interleaving is really only needed for the R2), then you will have daily temperature fluctuations affecting your data in a significant way. There is a publication on this effect by Wüthrich, I think. > Single diffusion tensor estimate: > I looked through the local tm model results and they appear to be all over > the place, although the relaxation behavior of the protein predicts that they > should be rather similar. > chi2 values were generally in the double to three (hundreds) digit region. > The model selection changes from residue to residue. > I don't believe that this sample dimerizes, but have not done a dilution > series yet. The local tm models are noisy - it is over-fitting - but the general trend in the data should explain the global diffusion tensor selected. Was the chi2 value high for the local tm models too? That would not good and might indicate that you have bad data. > R1, R2, NOE: > All fittings were done in Relax and heights and noise were measured following > recommendations in the relax manual. I also compared fits to fitting with > other methods. The errors and curve fits look ok. What about the chi2 values? Are these high? If not, that would again be a sign of the individual experiments being ok, but that temperature calibration and control might be an issue. > 13C Coupling: > I think I did not formulate this question clearly. My concern was with the > effect of dipolar couplings from 13C carbons on relaxation. > There is no question that the scalar coupling must be taken care of at the > experimental level and this is of course what I did. > Upon looking into this further, I believe that the effect from the dipolar > couplings should be negligible. That is what I understood - the 15N-13C dipolar relaxation. If you have fully labelled 15N and 13C samples, this is actually a significant source of relaxation. Significant in a model-free analysis would probably be a contribution of >5%. There are multiple publications on this issue, but unfortunately I cannot remember them off the top of my head. For example there was some work on this for analysing DNA. I know this because there is an abandoned branch of relax for handling multi-pole relaxation due to this exact effect in nucleic acids. Anyway, you should first determine if your data is good or bad. You could try the consistency testing analysis in relax for example. But first carefully check your temperature calibration and control. And also study the literature on 15N-13C dipolar relaxation in fully labelled samples. And, as I mentioned to Thibault, no one gets relaxation data measurements correct the first time ;) Regards, Edward _______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users
