Hi Paul, I am sending this to the relax-devel mailing to allow the discussions to be followed by others interested in relax development, and so that everyone including Remco knows what is happening. I have incorporated the data from Remco Sprangers that you sent to me (and which I attached to https://gna.org/task/?7712#comment6). The data is from: Remco Sprangers, Anna Gribun, Peter M. Hwang, Walid A. Houry, and Lewis E. Kay (2005) Quantitative NMR spectroscopy of supramolecular complexes: Dynamic side pores in ClpP are important for product release, PNAS, 102 (46), 16678-16683. (doi: http://dx.doi.org/10.1073/pnas.0507370102) The peak intensity and README files have been added to the test_suite/shared_data/dispersion/Spranger_ClpP/ directory in the relax_disp relax branch. The data will be used to test the implementation the 'MQ NS 2-site' dispersion model (the multi-quantum 2-site numerical solution) which will soon be added to relax from the Lescanne and Marion code attached to https://gna.org/task/?7712#comment7. Unfortunately I don't have enough detail to write a script to analyse this yet. Did Remco mention for which temperature this data is for? I would guess that this is the data shown in figure 3b. I will make the safe assumption that residues 135 and 137 correspond to I149 and I151 in the paper. The mapping of CPMG frequencies of 66.7 to 1,000 Hz mentioned in the methods section to the ncyc values of 12,1,10,2,8,3,6,4,5,1,2,3 in the README file does not match. But assuming figure 3b, the nuCPMG values go up to 800 Hz, which is 66.7 * 12. So I will assume 66.7 Hz (or 800/12) for one ncyc unit. Assuming the data is from figure 3b, I will then test that relax finds kex = 67.5 s^-1, pF = pS = 0.5, that the optimised delta_omega proton and carbon values match those in the README file, and that R20_S and R20_F for each residue and field are close to the plateau values in figure 3b. I will treat the data for states S and F as separate spins of the same residue, so that 4 independent spins are optimised with no assumptions about their state. The initial values for delta_omega can be used and a grid search performed over kex, pF, and the two R20_S and R20_F parameters - and then all parameters optimised together. This would then be equivalent to the methods section statement "Chemical shift differences were initially estimated from spectra recorded at 0.5°C and then optimized based on the extracted exchange parameters from fits of the dispersion data". Such an optimisation strategy would be only for dedicated relax scripts and would not be present in the auto-analysis presented in the GUI and current sample scripts. We should discuss this in a separate thread, as you have mentioned this idea before. Cheers, Edward