Identification and characterization of lysine-methylated sites on histones and non-histone proteins

Tzong Yi Lee, Cheng Wei Chang, Cheng Tzung Lu, Tzu Hsiu Cheng, Tzu Hao Chang

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

Protein methylation is a kind of post-translational modification (PTM), and typically takes place on lysine and arginine amino acid residues. Protein methylation is involved in many important biological processes, and most recent studies focused on lysine methylation of histones due to its critical roles in regulating transcriptional repression and activation. Histones possess highly conserved sequences and are homologous in most species. However, there is much less sequence conservation among non-histone proteins. Therefore, mechanisms for identifying lysine-methylated sites may greatly differ between histones and non-histone proteins. Nevertheless, this point of view was not considered in previous studies. Here we constructed two support vector machine (SVM) models by using lysine-methylated data from histones and non-histone proteins for predictions of lysine-methylated sites. Numerous features, such as the amino acid composition (AAC) and accessible surface area (ASA), were used in the SVM models, and the predictive performance was evaluated using five-fold cross-validations. For histones, the predictive sensitivity was 85.62% and specificity was 80.32%. For non-histone proteins, the predictive sensitivity was 69.1% and specificity was 88.72%. Results showed that our model significantly improved the predictive accuracy of histones compared to previous approaches. In addition, features of the flanking region of lysine-methylated sites on histones and non-histone proteins were also characterized and are discussed. A gene ontology functional analysis of lysine-methylated proteins and correlations of lysine-methylated sites with other PTMs in histones were also analyzed in detail. Finally, a web server, MethyK, was constructed to identify lysine-methylated sites.

Original languageEnglish
Pages (from-to)11-18
Number of pages8
JournalComputational Biology and Chemistry
Volume50
DOIs
Publication statusPublished - 2014

Fingerprint

Histones
Lysine
Proteins
Protein
Methylation
Specificity
Support vector machines
Amino Acids
Amino acids
Support Vector Machine
Pulse time modulation
Functional analysis
Arginine
Biological Phenomena
Gene Ontology
Web Server
Functional Analysis
Conserved Sequence
Surface area
Cross-validation

Keywords

  • Histone
  • Lysine
  • Methylation
  • Non-histone
  • Post-translational modification
  • PTM
  • Support vector machine
  • SVM

ASJC Scopus subject areas

  • Biochemistry
  • Structural Biology
  • Organic Chemistry
  • Computational Mathematics
  • Medicine(all)

Cite this

Identification and characterization of lysine-methylated sites on histones and non-histone proteins. / Lee, Tzong Yi; Chang, Cheng Wei; Lu, Cheng Tzung; Cheng, Tzu Hsiu; Chang, Tzu Hao.

In: Computational Biology and Chemistry, Vol. 50, 2014, p. 11-18.

Research output: Contribution to journalArticle

Lee, Tzong Yi ; Chang, Cheng Wei ; Lu, Cheng Tzung ; Cheng, Tzu Hsiu ; Chang, Tzu Hao. / Identification and characterization of lysine-methylated sites on histones and non-histone proteins. In: Computational Biology and Chemistry. 2014 ; Vol. 50. pp. 11-18.
@article{9e5ea55748e64d8ebb72b8d2d7c27851,
title = "Identification and characterization of lysine-methylated sites on histones and non-histone proteins",
abstract = "Protein methylation is a kind of post-translational modification (PTM), and typically takes place on lysine and arginine amino acid residues. Protein methylation is involved in many important biological processes, and most recent studies focused on lysine methylation of histones due to its critical roles in regulating transcriptional repression and activation. Histones possess highly conserved sequences and are homologous in most species. However, there is much less sequence conservation among non-histone proteins. Therefore, mechanisms for identifying lysine-methylated sites may greatly differ between histones and non-histone proteins. Nevertheless, this point of view was not considered in previous studies. Here we constructed two support vector machine (SVM) models by using lysine-methylated data from histones and non-histone proteins for predictions of lysine-methylated sites. Numerous features, such as the amino acid composition (AAC) and accessible surface area (ASA), were used in the SVM models, and the predictive performance was evaluated using five-fold cross-validations. For histones, the predictive sensitivity was 85.62{\%} and specificity was 80.32{\%}. For non-histone proteins, the predictive sensitivity was 69.1{\%} and specificity was 88.72{\%}. Results showed that our model significantly improved the predictive accuracy of histones compared to previous approaches. In addition, features of the flanking region of lysine-methylated sites on histones and non-histone proteins were also characterized and are discussed. A gene ontology functional analysis of lysine-methylated proteins and correlations of lysine-methylated sites with other PTMs in histones were also analyzed in detail. Finally, a web server, MethyK, was constructed to identify lysine-methylated sites.",
keywords = "Histone, Lysine, Methylation, Non-histone, Post-translational modification, PTM, Support vector machine, SVM",
author = "Lee, {Tzong Yi} and Chang, {Cheng Wei} and Lu, {Cheng Tzung} and Cheng, {Tzu Hsiu} and Chang, {Tzu Hao}",
year = "2014",
doi = "10.1016/j.compbiolchem.2014.01.009",
language = "English",
volume = "50",
pages = "11--18",
journal = "Computational Biology and Chemistry",
issn = "1476-9271",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Identification and characterization of lysine-methylated sites on histones and non-histone proteins

AU - Lee, Tzong Yi

AU - Chang, Cheng Wei

AU - Lu, Cheng Tzung

AU - Cheng, Tzu Hsiu

AU - Chang, Tzu Hao

PY - 2014

Y1 - 2014

N2 - Protein methylation is a kind of post-translational modification (PTM), and typically takes place on lysine and arginine amino acid residues. Protein methylation is involved in many important biological processes, and most recent studies focused on lysine methylation of histones due to its critical roles in regulating transcriptional repression and activation. Histones possess highly conserved sequences and are homologous in most species. However, there is much less sequence conservation among non-histone proteins. Therefore, mechanisms for identifying lysine-methylated sites may greatly differ between histones and non-histone proteins. Nevertheless, this point of view was not considered in previous studies. Here we constructed two support vector machine (SVM) models by using lysine-methylated data from histones and non-histone proteins for predictions of lysine-methylated sites. Numerous features, such as the amino acid composition (AAC) and accessible surface area (ASA), were used in the SVM models, and the predictive performance was evaluated using five-fold cross-validations. For histones, the predictive sensitivity was 85.62% and specificity was 80.32%. For non-histone proteins, the predictive sensitivity was 69.1% and specificity was 88.72%. Results showed that our model significantly improved the predictive accuracy of histones compared to previous approaches. In addition, features of the flanking region of lysine-methylated sites on histones and non-histone proteins were also characterized and are discussed. A gene ontology functional analysis of lysine-methylated proteins and correlations of lysine-methylated sites with other PTMs in histones were also analyzed in detail. Finally, a web server, MethyK, was constructed to identify lysine-methylated sites.

AB - Protein methylation is a kind of post-translational modification (PTM), and typically takes place on lysine and arginine amino acid residues. Protein methylation is involved in many important biological processes, and most recent studies focused on lysine methylation of histones due to its critical roles in regulating transcriptional repression and activation. Histones possess highly conserved sequences and are homologous in most species. However, there is much less sequence conservation among non-histone proteins. Therefore, mechanisms for identifying lysine-methylated sites may greatly differ between histones and non-histone proteins. Nevertheless, this point of view was not considered in previous studies. Here we constructed two support vector machine (SVM) models by using lysine-methylated data from histones and non-histone proteins for predictions of lysine-methylated sites. Numerous features, such as the amino acid composition (AAC) and accessible surface area (ASA), were used in the SVM models, and the predictive performance was evaluated using five-fold cross-validations. For histones, the predictive sensitivity was 85.62% and specificity was 80.32%. For non-histone proteins, the predictive sensitivity was 69.1% and specificity was 88.72%. Results showed that our model significantly improved the predictive accuracy of histones compared to previous approaches. In addition, features of the flanking region of lysine-methylated sites on histones and non-histone proteins were also characterized and are discussed. A gene ontology functional analysis of lysine-methylated proteins and correlations of lysine-methylated sites with other PTMs in histones were also analyzed in detail. Finally, a web server, MethyK, was constructed to identify lysine-methylated sites.

KW - Histone

KW - Lysine

KW - Methylation

KW - Non-histone

KW - Post-translational modification

KW - PTM

KW - Support vector machine

KW - SVM

UR - http://www.scopus.com/inward/record.url?scp=84902242497&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84902242497&partnerID=8YFLogxK

U2 - 10.1016/j.compbiolchem.2014.01.009

DO - 10.1016/j.compbiolchem.2014.01.009

M3 - Article

C2 - 24560580

AN - SCOPUS:84902242497

VL - 50

SP - 11

EP - 18

JO - Computational Biology and Chemistry

JF - Computational Biology and Chemistry

SN - 1476-9271

ER -