On Feb 8, 2008 4:54 PM, Sebastien Morin <sebastien.morin.1@xxxxxxxxx> wrote:
Hi, I have a theoretical question about Model-Free analysis concerning the minimum amount of data required... In ModelFree, it is possible to obtain fitted parameters for residues for which only one R1 and one R2 datasets are available (no NOE at all). What do you think of this ? Can these residues be fitted reliably to model 1 ? Or is this worthless to have data for only R1 and R2 and no NOE ?
Well, it's always possible to fit anything to anything. Note that the R2 best samples the spectral density at zero frequency J(0), the R1 best at the heteronuclear frequency J(wX), and the NOE at the proton frequency J(wH). So, this is why the R1 doesn't tell you much, it is a bit like a pivot point in the spectral density curves. If you have fast, low amplitude internal motions, J(0) is relatively high and J(wX) is relatively low. If you have slow, large amplitude motions then the reverse happens. J(wX) is close to the pivot of this. Note that timescale and amplitude, as well as the global tumbling, are all convoluted together causing these J(w) curves to change. So, in answer to your question, you can use the R1 and R2 if you have fast, low amplitude internal motions. If you have anything interesting happening though, then you desperately need the NOE.
Also, what happens when such data is input in relax ? Is the residue rejected ? What happens if a residue as a full dataset at one field (R1, R2 and NOE) and an incomplete dataset at another field (R1 and R2 only). Will the 5 measured information be used or will only the complete dataset from the first field be used and the incompelte dataset rejected ?
Everything will be accepted. The spin system will be skipped during optimisation only in a few cases. This occurs in the overfit_deselect() function for the different types of analysis. So for model-free analysis, this will happen if the spin is deselected, if there is zero relaxation data, and if the number of parameters of the model is greater than the number of relaxation data sets. One last condition is that if there is 2 or less relaxation data points, then the spin is deselected. This condition could be removed though, if anyone at anytime would like to use just the R1 and R2 at a single field. Regards, Edward