Electrostatic Similarity Determination Using Multiresolution Analysis

Huseyin Hakkoymaz; Chris A Kieslich; Ronald D Gorham Jr; Dimitrios Gunopulos; Dimitrios Morikis

doi:10.1002/minf.201100002

Electrostatic Similarity Determination Using Multiresolution Analysis

Mol Inform. 2011 Aug;30(8):733-46. doi: 10.1002/minf.201100002. Epub 2011 Aug 4.

Authors

Huseyin Hakkoymaz¹, Chris A Kieslich², Ronald D Gorham Jr², Dimitrios Gunopulos³, Dimitrios Morikis⁴

Affiliations

¹ Department of Computer Science and Engineering, University of California Riverside, Riverside, CA, 92521, USA.
² Department of Bioengineering, University of California Riverside, Riverside, CA, 92521, USA phone: +19518272696.
³ Department of Informatics and Telecommunications, University of Athens, Athens 15784, Greece.
⁴ Department of Bioengineering, University of California Riverside, Riverside, CA, 92521, USA phone: +19518272696. dmorikis@engr.ucr.edu.

PMID: 27467264
DOI: 10.1002/minf.201100002

Abstract

Molecular similarity is an important tool in protein and drug design for analyzing the quantitative relationships between physicochemical properties of two molecules. We present a family of similarity measures which exploits the ability of wavelet transformation to analyze the spectral components of physicochemical properties and suggests a sensitive way for measuring similarities of biological molecules. In order to investigate how effective wavelet-based similarity measures were against conventional measures, we defined several patterns which involve scalar or topological changes in the distribution of electrostatic properties. The wavelet-based measures were more successful in discriminating these patterns in contrast to the current state-of-art similarity measures. We also present the validity of wavelet-based similarity measures through the hierarchical clustering of two protein datasets consisting of families of homologous domains and alanine scan mutants. This type of similarity analysis is useful for protein structure-function studies and protein design.

Keywords: Electrostatics; Machine learning; Molecular similarity; Multiresolution analysis; Wavelet transformation.