Generation and evaluation of dimension-reduced amino acid parameter representations by artificial neural networks

Jens Meiler, Michael Müller, Anita Zeidler, Felix Schmäschke

Research output: Contribution to journalArticle

112 Citations (Scopus)

Abstract

In order to process data of proteins, a numerical representation for an amino acid is often necessary. Many suitable parameters can be derived from experiments or statistical analysis of databases. To ensure a fast and efficient use of these sources of information, a reduction and extraction of relevant information out of these parameters is a basic need. In this approach established methods like principal component analysis (PCA) are supplemented by a method based on symmetric neural networks. Two different parameter representations of amino acids are reduced from five and seven dimensions, respectively, to one, two, three, or four dimensions by using a symmetric neural network approach alternatively with one or three hidden layers. It is possible to create general reduced parameter representations for amino acids. To demonstrate the ability of this approach, these reduced sets of parameters are applied for the ab initio prediction of protein secondary structure from primary structure only. Artificial neural networks are implemented and trained with a diverse representation of 430 proteins out of the PDB. An essentially faster training and also prediction without a decrease in accuracy is obtained for the reduced parameter representations in comparison with the complete set of parameters. The method is transferable to other amino acids or even other molecular building blocks, like nucleic acids, and therefore represents a general approach.

Original languageEnglish
Pages (from-to)360-369
Number of pages10
JournalJournal of Molecular Modeling
Volume7
Issue number9
DOIs
Publication statusPublished - 1 Jan 2001

Keywords

  • Amino acid parameters
  • Neural networks
  • Quantitative structure-property relation
  • Secondary structure prediction

ASJC Scopus subject areas

  • Catalysis
  • Computer Science Applications
  • Physical and Theoretical Chemistry
  • Organic Chemistry
  • Computational Theory and Mathematics
  • Inorganic Chemistry

Cite this

Generation and evaluation of dimension-reduced amino acid parameter representations by artificial neural networks. / Meiler, Jens; Müller, Michael; Zeidler, Anita; Schmäschke, Felix.

In: Journal of Molecular Modeling, Vol. 7, No. 9, 01.01.2001, p. 360-369.

Research output: Contribution to journalArticle

@article{b6658b701ced46549cc1b67f6584405a,
title = "Generation and evaluation of dimension-reduced amino acid parameter representations by artificial neural networks",
abstract = "In order to process data of proteins, a numerical representation for an amino acid is often necessary. Many suitable parameters can be derived from experiments or statistical analysis of databases. To ensure a fast and efficient use of these sources of information, a reduction and extraction of relevant information out of these parameters is a basic need. In this approach established methods like principal component analysis (PCA) are supplemented by a method based on symmetric neural networks. Two different parameter representations of amino acids are reduced from five and seven dimensions, respectively, to one, two, three, or four dimensions by using a symmetric neural network approach alternatively with one or three hidden layers. It is possible to create general reduced parameter representations for amino acids. To demonstrate the ability of this approach, these reduced sets of parameters are applied for the ab initio prediction of protein secondary structure from primary structure only. Artificial neural networks are implemented and trained with a diverse representation of 430 proteins out of the PDB. An essentially faster training and also prediction without a decrease in accuracy is obtained for the reduced parameter representations in comparison with the complete set of parameters. The method is transferable to other amino acids or even other molecular building blocks, like nucleic acids, and therefore represents a general approach.",
keywords = "Amino acid parameters, Neural networks, Quantitative structure-property relation, Secondary structure prediction",
author = "Jens Meiler and Michael M{\"u}ller and Anita Zeidler and Felix Schm{\"a}schke",
year = "2001",
month = "1",
day = "1",
doi = "10.1007/s008940100038",
language = "English",
volume = "7",
pages = "360--369",
journal = "Journal of Molecular Modeling",
issn = "1610-2940",
publisher = "Springer Verlag",
number = "9",

}

TY - JOUR

T1 - Generation and evaluation of dimension-reduced amino acid parameter representations by artificial neural networks

AU - Meiler, Jens

AU - Müller, Michael

AU - Zeidler, Anita

AU - Schmäschke, Felix

PY - 2001/1/1

Y1 - 2001/1/1

N2 - In order to process data of proteins, a numerical representation for an amino acid is often necessary. Many suitable parameters can be derived from experiments or statistical analysis of databases. To ensure a fast and efficient use of these sources of information, a reduction and extraction of relevant information out of these parameters is a basic need. In this approach established methods like principal component analysis (PCA) are supplemented by a method based on symmetric neural networks. Two different parameter representations of amino acids are reduced from five and seven dimensions, respectively, to one, two, three, or four dimensions by using a symmetric neural network approach alternatively with one or three hidden layers. It is possible to create general reduced parameter representations for amino acids. To demonstrate the ability of this approach, these reduced sets of parameters are applied for the ab initio prediction of protein secondary structure from primary structure only. Artificial neural networks are implemented and trained with a diverse representation of 430 proteins out of the PDB. An essentially faster training and also prediction without a decrease in accuracy is obtained for the reduced parameter representations in comparison with the complete set of parameters. The method is transferable to other amino acids or even other molecular building blocks, like nucleic acids, and therefore represents a general approach.

AB - In order to process data of proteins, a numerical representation for an amino acid is often necessary. Many suitable parameters can be derived from experiments or statistical analysis of databases. To ensure a fast and efficient use of these sources of information, a reduction and extraction of relevant information out of these parameters is a basic need. In this approach established methods like principal component analysis (PCA) are supplemented by a method based on symmetric neural networks. Two different parameter representations of amino acids are reduced from five and seven dimensions, respectively, to one, two, three, or four dimensions by using a symmetric neural network approach alternatively with one or three hidden layers. It is possible to create general reduced parameter representations for amino acids. To demonstrate the ability of this approach, these reduced sets of parameters are applied for the ab initio prediction of protein secondary structure from primary structure only. Artificial neural networks are implemented and trained with a diverse representation of 430 proteins out of the PDB. An essentially faster training and also prediction without a decrease in accuracy is obtained for the reduced parameter representations in comparison with the complete set of parameters. The method is transferable to other amino acids or even other molecular building blocks, like nucleic acids, and therefore represents a general approach.

KW - Amino acid parameters

KW - Neural networks

KW - Quantitative structure-property relation

KW - Secondary structure prediction

UR - http://www.scopus.com/inward/record.url?scp=0035789525&partnerID=8YFLogxK

U2 - 10.1007/s008940100038

DO - 10.1007/s008940100038

M3 - Article

VL - 7

SP - 360

EP - 369

JO - Journal of Molecular Modeling

JF - Journal of Molecular Modeling

SN - 1610-2940

IS - 9

ER -