Inferential Structure Determination from NMR Data

  • Michael Habeck
Part of the Statistics for Biology and Health book series (SBH)


The standard approach to biomolecular structure calculation from nuclear magnetic resonance (NMR) data is to solve a non-linear optimization problem by simulated annealing or some other optimization method. Despite practical success, fundamental issues such as the definition of a meaningful coordinate error, the assessment of the goodness of fit to the data, and the estimation of missing parameter values remain unsolved. Inferential structure determination (ISD) is a principled alternative to optimization approaches. ISD applies Bayesian reasoning to represent the unknown molecular structure and its uncertainty through a posterior probability distribution. The posterior distribution also determines missing parameter values such as NMR alignment tensors and error parameters quantifying the quality of the data. The atomic coordinates and additional unknowns are estimated from their joint posterior probability distribution with a parallel Markov chain Monte Carlo algorithm.


Nuclear Magnetic Resonance Posterior Distribution Nuisance Parameter Nuclear Magnetic Resonance Data Hybrid Energy 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.



ISD has been developed in close collaboration with Wolfgang Rieping and Michael Nilges (Institut Pasteur, Paris). This work has been supported by Deutsche Forschungsgemeinschaft (DFG) grant HA 5918/1-1 and the Max Planck Society.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  1. 1.Department of Protein EvolutionMax-Planck-Institute for Developmental BiologyTübingenGermany
  2. 2.Department of Empirical InferenceMax-Planck-Institute for Intelligent SystemsTübingenGermany

Personalised recommendations