Modelfree4

From relax wiki
Revision as of 15:54, 15 December 2015 by Bugman (talk | contribs) (Added the Category:Software category.)
Jump to navigation Jump to search

Art Palmer's software Modelfree4 is designed for the model-free analysis for NMR relaxation data. Modelfree4 can be used as an optimisation engine to replace the minimisation algorithms implemented within relax.

Details

Website

The original Modelfree4 website is http://cumc.columbia.edu/dept/gsas/biochem//labs/palmer/software/modelfree.html.

Overview

From the Modelfree4 manual:

Modelfree (version 4.0) is a program to fit the extended model free spectral density function to NMR spin relaxation data. The program can analyze the spin-lattice relaxation rate constant (R1), the spin-spin relaxation rate constant (R2), and the heteronuclear steady-state {1H}-X nuclear Overhauser effect (NOE) for any combination of 13C and 15N spins at up to five static magnetic fields. The algorithm assumes dipolar and chemical shift anisotropy (CSA) relaxation mechanisms for R1, R2, and the NOE and includes an additive term to account for chemical exchange broadening of R2. The exchange broadening parameter is scaled quadratically with respect to the static magnetic field if data for more than one field is available. Descriptions of the main principles utilized in the program for model selection, optimization and error analysis using Monte Carlo simulations have been published previously (Palmer et al., 1991; Mandel et al., 1995).

Modelfree incorporates three models for rotational diffusion. Brent's method is used to optimize a single global τm for an isotropic overall diffusion model. Either Powell's method (as implemented by Brent) or a simulated annealing protocol (based on the downhill simplex method) is used to optimize a global rotational diffusion tensor for an axially symmetric diffusion model. A local rotational correlation time, τmi, for each spin can be optimized by non-linear least squares regression.

Other internal model free parameters (order parameters, internal correlation times and chemical exchange terms) are optimized by restrained non-linear least squares. Any parameter can be fixed at its input value rather than optimized and simple bounds can be placed on any parameter.

Monte Carlo simulations are used to estimate uncertainties in model-free parameters and to perform statistical model selection based on F-testing. A good introductory discussion of the use of Monte Carlo simulations in error analysis is given by Press et al. (Press et al., 1986).

Authors

  • Arthur G. Palmer, III.

Version

The last release was 4.20 from December 2006.

Reference

  • Mandel, A. M., Akke, M., and Palmer, 3rd, A. G. (1995). Backbone dynamics of Escherichia coli ribonuclease HI: correlations with structure and function in an active enzyme. J. Mol. Biol., 246(1), 144-163. (DOI: 10.1006/jmbi.1994.0073)
  • Palmer, 3rd, A. G., Rance, M., and Wright, P. E. (1991). Intramolecular motions of a zinc finger DNA-binding domain from Xfin characterized by proton-detected natural abundance carbon-13 heteronuclear NMR spectroscopy. J. Am. Chem. Soc., 113(12), 4371-4380. (DOI: 10.1021/ja00012a001)

Usage in relax

To use ModelFree4 within relax, the following user function are provided:

These allow Modelfree4 to be used as a model-free optimisation within relax, replacing the built-in algorithms.

See also