Fast Structural Alignment of Biomolecules Using a Hash Table, N-Grams and String Descriptors

Algorithms

Raphael André Bauer⁴  Kristian Rother¹  Peter Moor⁴  Knut Reinert³  Thomas Steinke²  Janusz M. Bujnicki¹ 

[1] International Institute of Molecular and Cell Biology in Warsaw, ul. Ks. Trojdena 4, 02-109 Warsaw, Poland;Zuse Institute Berlin, Dept. Computer Science, Takustrasse 7, 14195 Berlin, Germany;Freie Universitӓt Berlin, Algorithmische Bioinformatik, Institut fϋr Informatik, Takustr. 9, 14195 Berlin, Germany;Charité Medical University, Structural Bioinformatics Group, Arnimallee 22, 14195 Berlin, Germany

关键词: Structural alignment;    protein;    RNA;    hash table;    n-gram;    torsion angles;   

DOI  :  10.3390/a2020692

来源: mdpi

【 摘 要 】

This work presents a generalized approach for the fast structural alignment of thousands of macromolecular structures. The method uses string representations of a macromolecular structure and a hash table that stores n-grams of a certain size for searching. To this end, macromolecular structure-to-string translators were implemented for protein and RNA structures. A query against the index is performed in two hierarchical steps to unite speed and precision. In the first step the query structure is translated into n-grams, and all target structures containing these n-grams are retrieved from the hash table. In the second step all corresponding n-grams of the query and each target structure are subsequently aligned, and after each alignment a score is calculated based on the matching n-grams of query and target. The extendable framework enables the user to query and structurally align thousands of protein and RNA structures on a commodity machine and is available as open source from http://lajolla.sf.net.

【 授权许可】

CC BY   
© 2009 by the authors; licensee Molecular Diversity Preservation International, Basel, Switzerland.