Dear all, after some time now I installed the GUI version of relax4.1.2 on Ubuntu 18.04.2 LTS (with 16GB RAM and an i7 7700 CPU @3,6 GHz x8 processor) and repeated the R1 determination with the same result. After doing the grid search successfully the Monte Carlo simulations run for a few seconds until around simulation 20 and then get stuck there. Unfortunately I don't get an error message and I can still navigate between open relax windows so the program itself does not appear to have crashed. I tried this now on three different computers (2 Macs and 1 Ubuntu, everytime with the relax GUI) always getting the same result. As far as I can tell I did the setup as described in the manual but maybe I did a mistake there? I use the sparky format for the peak lists from single-scan interleaved spectra and read out the heights of the signals. Regarding the spins I select the N and H spins for the backbone amides, for error analysis I use two triplicated spectra.
Any ideas are highly appreciated, thanks in advance! Cheers, Oliver -- Oliver Arnolds, M.Sc. AG Biomolekulare NMR (NC5/173) Faculty of Chemistry and Biochemistry Ruhr-University Bochum Universitaetsstrasse 150 44801 Bochum Germany Tel.: +49 234 32-26246 Am 14.03.19, 17:41 schrieb "Oliver Arnolds via nmr-relax-users" <[email protected]>: Hi Edward, thanks for your quick reply! Yes, I tried it already on another Mac with better hardware, but unfortunately it also didn't work. Telling you what exactly I was doing, would have been of course very helpful, sorry about that. I was just running the determination of the relaxation rates, I didn't even get to the model-free analysis. Determination of R1 and R2 showed the same problem. This is what surprised me because I did not expect relax to use so much CPU if it's not a model-free or relaxation dispersion analysis. I used temperature compensation blocks as well as single scan interleaving, so I hope that this won't cause any issues once I start the model-free analysis. I will try to install the Linux distribution on a more powerful machine with Ubuntu and start another run in the meantime. Have a nice holiday! Regards, Oliver -- Oliver Arnolds, M.Sc. AG Biomolekulare NMR (NC5/173) Faculty of Chemistry and Biochemistry Ruhr-University Bochum Universitaetsstrasse 150 44801 Bochum Germany Tel.: +49 234 32-26246 Am 14.03.19, 10:20 schrieb "Edward d'Auvergne" <[email protected]>: On Thu, 14 Mar 2019 at 10:00, Oliver Arnolds <[email protected]> wrote: > > Dear all, > > I am new to using relax and run the GUI of relax 4.0.8. and 4.1.1 version on Mac OS High Sierra 10.13.6. When trying to fit my relaxation data I ran into to following problem: The Monte Carlo simulation seemed to get stuck at random simulation numbers (mostly before reaching 100) without giving an error output. I thought it might have been a hardware problem as 100% of the CPU was used so I reduced the number of spins to 20 and it worked. By coincidence I found that after successfully running a MC with a small number of spins and then repeating it with the whole protein strangely seemed to solve the problem. As I was able to reproduce this I set up a new pipeline for another protein of similar size and repeated the procedure as before which then failed. When going back to the pipeline that already worked before there is no issue when running relax for the whole protein. > I encounter this problem for both versions of relax that I use and as far as I can tell I didn’t make a mistake regarding the setup. I tried it on another Mac with better hardware and it showed the same result, too. > Did anyone else ran into this problem before or does know a solution for it? I already searched the archive but did not find any post that described something similar. > I hope you can help me! Hi Oliver, Welcome to the relax users mailing list! Please note that I will be going on a 3 week holiday starting tomorrow, so I may not be able to help you so quickly. The issue as you describe it does not sound like a software issue but, as you suggest, a hardware issue. Can you run this on a different computer to check? If that works, re-run it on the original computer. Could you also give more details about what you are running? There are many NMR dynamics analysis types implemented in relax. I can only assume that you are running the automated model-free analysis in the GUI. If so, which part of the analysis do the MC simulations fail on? Note that some simulations can take a long, long time. Especially if there is some bad input data or an inaccurate 3D structure is used for the vector orientations. Also note that this is often the case in model-free analyses if the essential temperature control and calibration has not been performed: http://www.nmr-relax.com/manual/Temperature_control_and_calibration.html I hope some of this scattered information helps. Regards, Edward _______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users _______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users _______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users
