Hi, You have to be careful with error analysis as the errors are as important as the data itself! By 4 sets of R1 values, I would assume that you have these all at the same field strength. With this data there are a number of ways the data can be used in relax and a number of ways its error can be calculated. For the error calculation there are a number of methods that can be used. For example for a single R1 experiment, either the RMSD of the base plane noise or duplicate spectra can be used. This allows you to calculated an average error for all peaks in a single spectra (which can then be averaged over all spectra if not all time points are duplicated). Alternatively with triplicate spectra you can calculate one error value per peak per spectrum. Both approaches have good and bad points. The single average value for all peaks per spectrum suffers from bias as not all peaks are affected by the noise in the same way (bigger peaks are less affected). The single value for all peaks for all spectra (averaging previous values), introduces a second bias as the noise tends to slightly decrease, exponentially, across the time points (R2 is most affected by this). Whereas the triplicate spectra approach with one error per peak per spectrum is bias free but suffers from being noisy. These errors are then propagated to the R1 error through Monte Carlo simulation. So in your case you can apply a different approach. You can calculate 4 R1 values for all spins you have data for, and ignore the normal error analysis approaches. Then you can calculate a mean and error from those 4 values. This error cannot be averaged across all spins though. In the previous paragraph this was possible because the error was for a single time point in the R1 curve, whereas this error is for the R1 value itself which will be different for each spin system. If temperature calibration has been done correctly, then this error per spin system should be bias free, although the error estimate will be noisy! I hope this helps. Regards, Edward P.S. Oh, I also have to warn you about loading 4 separate R1 data sets into one analysis - you will bias model-free analysis by given too much weight to the R1 values. And these are not the only methods you can use for error analysis. On 9/3/07, Sebastien Morin <sebastien.morin.1@xxxxxxxxx> wrote:
Hi ! I recorded 4 sets of R1 and would like to use them all and, so, extract a mean value and also an associated error... I would like to get the opinion of someone maybe more used with statistics than me... I thought about : 1. calculating the mean error 2. calculating the standard error (should be the best way, no) 3. calculating the standard deviation 4. extracting an error by calculating the extremes the value can reach in every dataset based on the error of each dataset What would the best error to use in a statistical point of view, but also in a model-free point of view..? Also, is there a way to use both the errors in the datasets and a error extrated for the observed deviation of data..? Note that the errors from each datasets were calculated directly from the fits, here using the 'autoFit.tcl' script from NMRPipe with data processed as Gaussian lines. Also, in the case of duplicates or triplicates, should one use the same approcah ? Thanks ! Séb :) -- ______________________________________ _______________________________________________ | | || Sebastien Morin || ||| Etudiant au PhD en biochimie ||| |||| Laboratoire de resonance magnetique nucleaire |||| ||||| Dr Stephane Gagne ||||| |||| CREFSIP (Universite Laval, Quebec, CANADA) |||| ||| 1-418-656-2131 #4530 ||| || || |_______________________________________________| ______________________________________ _______________________________________________ relax (http://nmr-relax.com) This is the relax-users mailing list relax-users@xxxxxxx To unsubscribe from this list, get a password reminder, or change your subscription options, visit the list information page at https://mail.gna.org/listinfo/relax-users