Normal Mode Flexible Fitting of High-Resolution Structure of Biological Molecules toward 1 Dimensional Low-Resolution Data

¹ University of Arizona

^* To whom correspondence should be addressed. E-mail: ftama{at}u.arizona.edu.

Submitted on September 18, 2007
Revised on October 1, 2007
Accepted on 17 October 2007

	Abstract

We present a method to reconstruct a three-dimensional structure from a pair distribution function by flexibly fitting known x-ray structures toward a conformation that agrees with the low-resolution data. This method uses a linear combination of low-frequency normal modes from elastic network description of the molecule in an iterative manner to deform the structure optimally to conform to the target pair distribution function. A simple function, pair distance distribution function between atoms, is chosen as a test model to establish computational algorithms - optimization algorithm and scoring function - that can utilize low resolution 1 dimensional data. In order to select a correct structural model based on less information, we developed a scoring function that takes into account a characteristic of pair distribution functions. In addition, we employ a new optimization algorithm, the trusted region method that relies on both first and second derivative of the scoring function. Illustrative results of our studies on simulated 1D data from five different proteins, for which large conformational changes are known to occur, are presented.

Key Words: coarse-grained model, conformational change, modeling, pair distribution function, protein, trusted region method