On Mon, 1 Apr 2019 at 16:26, Viennet, Thibault <[email protected]> wrote: > Hi, > > I have a question regarding how error bars are calculated in the CPMG-RD > fitting script. > > I got data on two very similar samples, both of which gave nice spectra with > similar S/N ratios. I ran cpmg_analysis scripts both with 500 Monte Carlo > simulations. > > However, one of them gave much higher error bars in the fits, leading to no > residues being selected I guess due the fact that error bars for the first > and last points overlap. Even reducing the insignificance value wouldn’t > solve this. > > What strikes me the most is the clear different behavior between the two set > of data, which look very similar from the spectroscopic point of view. Do you > have an idea about that?
Hi Thibault, Welcome to the relax mailing lists! Sorry for not responding earlier, I have just returned from holidays. As for the issue you are seeing, have you plotted your dispersion curves? You can use relax to generate plots of the dispersion curves with error bars, as well as base exponential curves (if you have collected such data). You can then use XMGrace to visualise all the curves together. See the relax_disp.plot_disp_curves and relax_disp.plot_exp_curves user functions for details. These plots should show you what is happening and why the errors are higher in one than the other. My guess is that either that the curves in one of the samples are much noisier than the other, or that the peak intensity errors have been incorrectly calculated. Both cases should be obvious from the plots, even if you cannot see it in the spectra. Regards, Edward _______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users
