In silico prediction of peptide binding affinity to class I mouse major histocompatibility complexes: A comparative molecular similarity index analysis (CoMSIA) study

Channa K. Hattotuwagama, Irini A. Doytchinova, DR Flower

Research output: Contribution to journalArticle

Abstract

Current methods for the in silico identification of T cell epitopes (which form the basis of many vaccines, diagnostics, and reagents) rely on the accurate prediction of peptide-major histocompatibility complex (MHC) affinity. A three-dimensional quantitative structure-activity relationship (3D-QSAR) for the prediction of peptide binding to class I MHC molecules was established using the comparative molecular similarity index analysis (CoMSIA) method. Three MHC alleles were studied: H2-Db,H2-Kb, and H2-Kk. Models were produced for each allele. Each model consisted of five physicochemical descriptors-steric bulk,
electrostatic potentials, hydrophobic interactions, and hydrogen-bond donor and hydrogen-bond acceptor abilities. The models have an acceptable level of predictivity: cross-validation leave-one-out statistical terms q2 and SEP (standard error of prediction) ranged between 0.490 and 0.679 and between 0.525 and 0.889, respectively. The non-cross-validated statistical terms r2
and SEE (standard error of estimate) ranged between 0.913 and 0.979 and between 0.167 and 0.248, respectively. The use of coefficient contour maps, which indicate favored and disfavored areas for each position of the MHC-bound peptides, allowed the binding specificity of each allele to be identified, visualized, and understood. The present study demonstrates the
effectiveness of CoMSIA as a method for studying peptide-MHC interactions. The peptides used in this study are available on the Internet (http://www.jenner.ac.uk/AntiJen). The partial least-squares method is available commercially in the SYBYL molecular modeling software package
Original languageEnglish
Pages (from-to)1415-1423
Number of pages9
JournalJournal of Chemical Information and Modeling
Volume45
Issue number5
DOIs
Publication statusPublished - Sep 2005

Fingerprint

Peptides
Hydrogen bonds
activity structure
Epitopes
interaction
T-Lymphocyte Epitopes
Molecular modeling
Vaccines
T-cells
diagnostic
Software packages
Electrostatics
Internet
ability
Molecules

Cite this

@article{ee2d1153075b4a0ab17a43d13e278067,
title = "In silico prediction of peptide binding affinity to class I mouse major histocompatibility complexes: A comparative molecular similarity index analysis (CoMSIA) study",
abstract = "Current methods for the in silico identification of T cell epitopes (which form the basis of many vaccines, diagnostics, and reagents) rely on the accurate prediction of peptide-major histocompatibility complex (MHC) affinity. A three-dimensional quantitative structure-activity relationship (3D-QSAR) for the prediction of peptide binding to class I MHC molecules was established using the comparative molecular similarity index analysis (CoMSIA) method. Three MHC alleles were studied: H2-Db,H2-Kb, and H2-Kk. Models were produced for each allele. Each model consisted of five physicochemical descriptors-steric bulk,electrostatic potentials, hydrophobic interactions, and hydrogen-bond donor and hydrogen-bond acceptor abilities. The models have an acceptable level of predictivity: cross-validation leave-one-out statistical terms q2 and SEP (standard error of prediction) ranged between 0.490 and 0.679 and between 0.525 and 0.889, respectively. The non-cross-validated statistical terms r2and SEE (standard error of estimate) ranged between 0.913 and 0.979 and between 0.167 and 0.248, respectively. The use of coefficient contour maps, which indicate favored and disfavored areas for each position of the MHC-bound peptides, allowed the binding specificity of each allele to be identified, visualized, and understood. The present study demonstrates theeffectiveness of CoMSIA as a method for studying peptide-MHC interactions. The peptides used in this study are available on the Internet (http://www.jenner.ac.uk/AntiJen). The partial least-squares method is available commercially in the SYBYL molecular modeling software package",
author = "Hattotuwagama, {Channa K.} and Doytchinova, {Irini A.} and DR Flower",
year = "2005",
month = "9",
doi = "10.1021/ci049667l",
language = "English",
volume = "45",
pages = "1415--1423",
journal = "Journal of Chemical Information and Modeling",
issn = "1549-9596",
publisher = "American Chemical Society",
number = "5",

}

In silico prediction of peptide binding affinity to class I mouse major histocompatibility complexes : A comparative molecular similarity index analysis (CoMSIA) study. / Hattotuwagama, Channa K.; Doytchinova, Irini A.; Flower, DR.

In: Journal of Chemical Information and Modeling, Vol. 45, No. 5, 09.2005, p. 1415-1423.

Research output: Contribution to journalArticle

TY - JOUR

T1 - In silico prediction of peptide binding affinity to class I mouse major histocompatibility complexes

T2 - A comparative molecular similarity index analysis (CoMSIA) study

AU - Hattotuwagama, Channa K.

AU - Doytchinova, Irini A.

AU - Flower, DR

PY - 2005/9

Y1 - 2005/9

N2 - Current methods for the in silico identification of T cell epitopes (which form the basis of many vaccines, diagnostics, and reagents) rely on the accurate prediction of peptide-major histocompatibility complex (MHC) affinity. A three-dimensional quantitative structure-activity relationship (3D-QSAR) for the prediction of peptide binding to class I MHC molecules was established using the comparative molecular similarity index analysis (CoMSIA) method. Three MHC alleles were studied: H2-Db,H2-Kb, and H2-Kk. Models were produced for each allele. Each model consisted of five physicochemical descriptors-steric bulk,electrostatic potentials, hydrophobic interactions, and hydrogen-bond donor and hydrogen-bond acceptor abilities. The models have an acceptable level of predictivity: cross-validation leave-one-out statistical terms q2 and SEP (standard error of prediction) ranged between 0.490 and 0.679 and between 0.525 and 0.889, respectively. The non-cross-validated statistical terms r2and SEE (standard error of estimate) ranged between 0.913 and 0.979 and between 0.167 and 0.248, respectively. The use of coefficient contour maps, which indicate favored and disfavored areas for each position of the MHC-bound peptides, allowed the binding specificity of each allele to be identified, visualized, and understood. The present study demonstrates theeffectiveness of CoMSIA as a method for studying peptide-MHC interactions. The peptides used in this study are available on the Internet (http://www.jenner.ac.uk/AntiJen). The partial least-squares method is available commercially in the SYBYL molecular modeling software package

AB - Current methods for the in silico identification of T cell epitopes (which form the basis of many vaccines, diagnostics, and reagents) rely on the accurate prediction of peptide-major histocompatibility complex (MHC) affinity. A three-dimensional quantitative structure-activity relationship (3D-QSAR) for the prediction of peptide binding to class I MHC molecules was established using the comparative molecular similarity index analysis (CoMSIA) method. Three MHC alleles were studied: H2-Db,H2-Kb, and H2-Kk. Models were produced for each allele. Each model consisted of five physicochemical descriptors-steric bulk,electrostatic potentials, hydrophobic interactions, and hydrogen-bond donor and hydrogen-bond acceptor abilities. The models have an acceptable level of predictivity: cross-validation leave-one-out statistical terms q2 and SEP (standard error of prediction) ranged between 0.490 and 0.679 and between 0.525 and 0.889, respectively. The non-cross-validated statistical terms r2and SEE (standard error of estimate) ranged between 0.913 and 0.979 and between 0.167 and 0.248, respectively. The use of coefficient contour maps, which indicate favored and disfavored areas for each position of the MHC-bound peptides, allowed the binding specificity of each allele to be identified, visualized, and understood. The present study demonstrates theeffectiveness of CoMSIA as a method for studying peptide-MHC interactions. The peptides used in this study are available on the Internet (http://www.jenner.ac.uk/AntiJen). The partial least-squares method is available commercially in the SYBYL molecular modeling software package

UR - https://pubs.acs.org/doi/10.1021/ci049667l

U2 - 10.1021/ci049667l

DO - 10.1021/ci049667l

M3 - Article

VL - 45

SP - 1415

EP - 1423

JO - Journal of Chemical Information and Modeling

JF - Journal of Chemical Information and Modeling

SN - 1549-9596

IS - 5

ER -