| Abstract
|
Modeling the structure of biological molec … Modeling the structure of biological molecules is critical for understandinghow these structures perform their function, and for designing compounds tomodify or enhance this function (for medicinal or industrial purposes). Thedetermination of molecular structure involves defining three-dimensionalpositions for each of the constituent atoms using a variety of experimental,theoretical and empirical data sources. Unfortunately, each of these datasources can be noisy or not available in sufficient abundance to determinethe precise position of each atom. Instead, some atomic positions areprecisely defined by the data, and others are poorly defined. Anunderstanding of structural uncertainty is critical for properly interpretingstructural models. We have developed a Bayesian approach for determining thecoordinates of atoms in a three-dimensional space. Our algorithm takes asinput a set of probabilistic constraints on the coordinates of the atoms, andan a priori distribution for each atom location. The output is a maximum aposteriori (MAP) estimate of the location of each atom. We introduceconstraints as updates to the prior distributions. In this paper, we describethe algorithm and show its performance on three data sets. The first dataset is synthetic and illustrates the convergence properties of the method. The other data sets comprise real biological data for a protein (the trprepressor molecule) and a nucleic acid (the transfer RNA fold). Finally, wedescribe how we have begun to extend the algorithm to make it suitable fornon-Gaussian constraints. e it suitable fornon-Gaussian constraints.
|
