Bayesian hidden Markov Model for DNA segmentation : A prior sensitivity analysis

Darfiana Nur; David Allingham; Judith Rousseau; Kerrie Mengersen; Ross Mcvinish

hal-00328181, version 1

Bayesian hidden Markov Model for DNA segmentation : A prior sensitivity analysis

Darfiana Nur ( Auteur à contacter de préférence ) ¹, David Allingham ², Judith Rousseau () ^3, 4, Kerrie Mengersen ⁵, Ross Mcvinish ⁵

Computational Statistics & Data Analysis (2009) 9999

Résumé : The focus of this paper is on the sensitivity to the specification of the prior in a hidden Markov model describing homogeneous segments of DNA sequences. An intron from the chimpanzee α-fetoprotein gene, which plays an im- portant role in embryonic development in mammals is analysed. Three main aims are considered : (i) to assess the sensitivity to prior specification in Bayesian hidden Markov models for DNA sequence segmentation; (ii) to examine the impact of replacing the standard Dirichlet prior with a mixture Dirichlet prior; and (iii) to propose and illus- trate a more comprehensive approach to sensitivity analysis, using importance sampling. It is obtained that (i) the posterior estimates obtained under a Bayesian hidden Markov model are indeed sensitive to the specification of the prior distributions; (ii) compared with the standard Dirichlet prior, the mixture Dirichlet prior is more flexible, less sensitive to the choice of hyperparameters and less constraining in the analysis, thus improving posterior estimates; and (iii) importance sampling was computationally feasible, fast and effective in allowing a richer sensitivity analysis.

1 : School of Mathematical and physical Sciences
University of Newcastle
2 : ARC center of excellence for Complex Dynamic Systems and Control
ARC center of excellence for complex dunamic systems and control
3 : CEntre de REcherches en MAthématiques de la DEcision (CEREMADE)
CNRS : UMR7534 – Université Paris IX - Paris Dauphine
4 : Centre de Recherche en Économie et Statistique (CREST)
INSEE – École Nationale de la Statistique et de l'Administration Économique
5 : school of mathematical sciences
Queensland University of Technology

Collaboration : Darfiana Nur,David Allingham, Judith Rousseau , Kerrie L. Mengersen, Ross McVinishd

hal-00328181, version 1
http://hal.archives-ouvertes.fr/hal-00328181
oai:hal.archives-ouvertes.fr:hal-00328181
Contributeur : Judith Rousseau <>
Soumis le : Vendredi 10 Octobre 2008, 22:10:25
Dernière modification le : Samedi 14 Février 2009, 17:00:07

Voir la fiche détaillée

Documents associés

PDF :

DOI : 10.1016/j.csda.2008.07.007

Exporter

Bibtex EndNote TEI RefWorks

%% hal-00328181, version 1
%% http://hal.archives-ouvertes.fr/hal-00328181
@article{nur:hal-00328181,
    hal_id = {hal-00328181},
    url = {http://hal.archives-ouvertes.fr/hal-00328181},
    title = {{Bayesian hidden Markov Model for DNA segmentation : A prior sensitivity analysis}},
    author = {Nur, Darfiana and Allingham, David and Rousseau, Judith and Mengersen, Kerrie and Mcvinish, Ross},
    abstract = {{The focus of this paper is on the sensitivity to the specification of the prior in a hidden Markov model describing homogeneous segments of DNA sequences. An intron from the chimpanzee $\alpha$-fetoprotein gene, which plays an im- portant role in embryonic development in mammals is analysed. Three main aims are considered : (i) to assess the sensitivity to prior specification in Bayesian hidden Markov models for DNA sequence segmentation; (ii) to examine the impact of replacing the standard Dirichlet prior with a mixture Dirichlet prior; and (iii) to propose and illus- trate a more comprehensive approach to sensitivity analysis, using importance sampling. It is obtained that (i) the posterior estimates obtained under a Bayesian hidden Markov model are indeed sensitive to the specification of the prior distributions; (ii) compared with the standard Dirichlet prior, the mixture Dirichlet prior is more flexible, less sensitive to the choice of hyperparameters and less constraining in the analysis, thus improving posterior estimates; and (iii) importance sampling was computationally feasible, fast and effective in allowing a richer sensitivity analysis.}},
    keywords = {DNA sequence; hidden Markov model; Bayesian model; sensitivity analysis; $\alpha$-fetoprotein; Markov chain Monte Carlo; importance sampling.},
    language = {Anglais},
    affiliation = {School of Mathematical and physical Sciences , ARC center of excellence for Complex Dynamic Systems and Control , CEntre de REcherches en MAth{\'e}matiques de la DEcision - CEREMADE , Centre de Recherche en {\'E}conomie et Statistique - CREST , school of mathematical sciences},
    pages = {9999},
    journal = {Computational Statistics \& Data Analysis},
    audience = {internationale },
    collaboration = {Darfiana Nur,David Allingham, Judith Rousseau , Kerrie L. Mengersen, Ross McVinishd },
    doi = {10.1016/j.csda.2008.07.007 },
    year = {2009},
    pdf = {http://hal.archives-ouvertes.fr/hal-00328181/PDF/intron\_CSDA\_edited.pdf},
}

%F hal-00328181, version 1
%0 Journal Article
%U http://hal.archives-ouvertes.fr/hal-00328181
%2 MATH:MATH_ST;STAT:TH
%T Bayesian hidden Markov Model for DNA segmentation : A prior sensitivity analysis
%A Nur, Darfiana
%A Allingham, David
%A Rousseau, Judith
%A Mengersen, Kerrie
%A Mcvinish, Ross
%+ School of Mathematical and physical Sciences , ARC center of excellence for Complex Dynamic Systems and Control , CEntre de REcherches en MAthématiques de la DEcision - CEREMADE , Centre de Recherche en Économie et Statistique - CREST , school of mathematical sciences
%X The focus of this paper is on the sensitivity to the specification of the prior in a hidden Markov model describing homogeneous segments of DNA sequences. An intron from the chimpanzee α-fetoprotein gene, which plays an im- portant role in embryonic development in mammals is analysed. Three main aims are considered : (i) to assess the sensitivity to prior specification in Bayesian hidden Markov models for DNA sequence segmentation; (ii) to examine the impact of replacing the standard Dirichlet prior with a mixture Dirichlet prior; and (iii) to propose and illus- trate a more comprehensive approach to sensitivity analysis, using importance sampling. It is obtained that (i) the posterior estimates obtained under a Bayesian hidden Markov model are indeed sensitive to the specification of the prior distributions; (ii) compared with the standard Dirichlet prior, the mixture Dirichlet prior is more flexible, less sensitive to the choice of hyperparameters and less constraining in the analysis, thus improving posterior estimates; and (iii) importance sampling was computationally feasible, fast and effective in allowing a richer sensitivity analysis.
%G Anglais
%2 2008
%R 10.1016/j.csda.2008.07.007
%J Computational Statistics & Data Analysis
%2 internationale
%8 2009-00-00
%2 2008
%P 9999
%K DNA sequence; hidden Markov model; Bayesian model; sensitivity analysis; α-fetoprotein; Markov chain Monte Carlo; importance sampling.
%Z Darfiana Nur,David Allingham, Judith Rousseau , Kerrie L. Mengersen, Ross McVinishd

<?xml version="1.0" encoding="UTF-8"?>
<listBibl xmlns="http://www.tei-c.org/ns/1.0" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML">
<listBibl type="article"><head>Articles dans des revues avec comité de lecture</head>
<biblStruct type="article" xml:id="hal-00328181"><analytic>
<author><persName><forename>Darfiana</forename><surname>Nur</surname></persName>
<affiliation>
<orgName type="team">School of Mathematical and physical Sciences</orgName>
<country>AU</country>
<orgName type="institution">University of Newcastle</orgName>
</affiliation>
</author>
<author><persName><forename>David</forename><surname>Allingham</surname></persName>
<affiliation>
<orgName type="team">ARC center of excellence for Complex Dynamic Systems and Control</orgName>
<country>AU</country>
<orgName type="institution">ARC center of excellence for complex dunamic systems and control</orgName>
</affiliation>
</author>
<author><persName><forename>Judith</forename><surname>Rousseau</surname></persName>
<affiliation>
<orgName type="laboratory">CEREMADE</orgName>
<orgName type="team">CEntre de REcherches en MAthématiques de la DEcision</orgName>
<country>FR</country>
<orgName type="institution">CNRS : UMR7534</orgName>
<orgName type="institution">Université Paris IX - Paris Dauphine</orgName>
</affiliation>
<affiliation>
<orgName type="laboratory">CREST</orgName>
<orgName type="team">Centre de Recherche en Économie et Statistique</orgName>
<country>FR</country>
<orgName type="institution">INSEE</orgName>
<orgName type="institution">École Nationale de la Statistique et de l'Administration Économique</orgName>
</affiliation>
</author>
<author><persName><forename>Kerrie</forename><surname>Mengersen</surname></persName>
<affiliation>
<orgName type="team">school of mathematical sciences</orgName>
<country>AU</country>
<orgName type="institution">Queensland University of Technology</orgName>
</affiliation>
</author>
<author><persName><forename>Ross</forename><surname>Mcvinish</surname></persName>
<affiliation>
<orgName type="team">school of mathematical sciences</orgName>
<country>AU</country>
<orgName type="institution">Queensland University of Technology</orgName>
</affiliation>
</author>
<title type="main" level="a">Bayesian hidden Markov Model for DNA segmentation : A prior sensitivity analysis</title></analytic><monogr><title level="j" type="main">Computational Statistics & Data Analysis</title><imprint><biblScope type="publicationDate">2009-00-00</biblScope><biblScope type="pp">9999</biblScope></imprint></monogr><idno type="doi">10.1016/j.csda.2008.07.007</idno><idno type="HALid">http://hal.archives-ouvertes.fr/hal-00328181</idno><idno type="HALFile">http://hal.archives-ouvertes.fr/hal-00328181/PDF/intron_CSDA_edited.pdf</idno></biblStruct>
</listBibl>
</listBibl>