This follows on from the post of:
http://thread.gmane.org/gmane.science.nmr.relax.devel/3882
Now in relax, the 'No Rex', 'LM63', and 'CR72' models for CPMG-type
experiments - respectively no chemical exchange, the Luz Meiboom 1963,
and the Carver and Richards 1972 models - can be optimised
successfully. Below is a comparison of the results from relax, NESSY,
CPMGFit, and ShereKhan. The NESSY chi-squared values are smaller most
likely due to a difference in how peak intensity errors are handled
within the program. To note: relax and ShereKhan, at least for these
models, consistently produce the same results with slightly better
accuracy than CPMGFit.
Regards,
Edward
The no exchange model
=====================
Residue :70
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 10.563 11.400 - -
R2 (800) 16.141 16.823 - -
chi2 3427.907 860.148 - -
Residue :71
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 5.830 - - -
chi2 47.357 - - -
The LM63 model
==============
Residue :70
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 6.806 7.639 6.866 6.806
R2 (800) 6.679 7.387 6.866 6.679
phi 0.310 0.259 1.510e-04 0.310
kex 4763.249 3906.180 2.342 4763.197
tau 4.199e-05 5.120e-05 0.427 4.199e-05
chi2 106.393 78.223 106.471 106.393
Residue :71
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 4.926 - 4.926 4.926
phi 0.059 - 2.963e-05 0.059
kex 2544.645 - 1.271 2544.636
tau 7.860e-04 - 0.786 7.860e-04
chi2 1.442 - 1.442 1.442
The CR72 model
==============
Residue :70
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 7.011 7.236 6.263 7.011
R2 (800) 9.465 10.027 6.263 9.465
pA 0.990 0.989 0.978 0.990
dw 5.577 5.373 0.937 5.577
kex 1765.993 1522.059 1.774 1765.988
chi2 18.450 7.701 65.237 18.450
Residue :71
-----------
Param relax NESSY CPMGFit ShereKhan
R2 (500) 4.978 - 4.983 4.978
pA 0.997 - 0.994 0.997
dw 4.460 - 0.757 4.461
kex 1879.584 - 0.896 1879.618
chi2 1.379 - 1.382 1.379
More comparisons of NMR relaxation dispersion software.