PDBsum entry 1zbj

Transferase

PDB id

1zbj

Contents

			Protein chain

					59 a.a.

References listed in PDB file

Key reference

Title

Inferential structure determination.

Authors

W.Rieping, M.Habeck, M.Nilges.

Ref.

Science, 2005, 309, 303-306. [DOI no: 10.1126/science.1110428]

PubMed id

16002620

Abstract

Macromolecular structures calculated from nuclear magnetic resonance data are not fully determined by experimental data but depend on subjective choices in data treatment and parameter settings. This makes it difficult to objectively judge the precision of the structures. We used Bayesian inference to derive a probability distribution that represents the unknown structure and its precision. This probability distribution also determines additional unknowns, such as theory parameters, that previously had to be chosen empirically. We implemented this approach by using Markov chain Monte Carlo techniques. Our method provides an objective figure of merit and improves structural quality.



	Figure 1. Fig. 1. Replica-exchange Monte Carlo algorithm. (A) We generate a stochastic sample (X(k), (k), (k)) from the joint posterior distribution in an iterative fashion by using Gibbs sampling (20). The nuisance parameters and are consecutively drawn from their conditional posterior distributions, with the values of the other parameters being fixed to their previously generated values. Coordinates are updated by using the hybrid Monte Carlo method (21). (B) To overcome energy barriers, we embed this scheme in a replica-exchange strategy, which simulates a sequence of heated copies of the system. Samples of the target distribution are generated in the low-temperature copy (T[low]) and propagate via stochastic exchanges between intermediate copies (T[low] < T[i] < T[high]) to the high-temperature system (T[high]). The temperature T[high] is chosen such that the polypeptide chain can move freely in order to escape local modes of the probability density.		Figure 3. Fig. 3. Estimation of nuisance parameters. Posterior histograms compiled from MCMC samples for the scaling factor in the ISPA and for the width of the log normal error distribution. (A) Posterior histogram p( - \|D,I) for the inverse sixth power of . This factor corrects interproton distances to match the experimental distances best. (B) Posterior histogram p( \|D,I) for the error . In conventional approaches, this analog to the weight (w[data] -2) can only be estimated via cross-validation or must be set empirically.