Modelo com erros de classificação para a proporção de não- disjunção cromossômica na meiose I

• Ano de defesa: 2007
• Autor(a) principal: Cristiane Silva Souto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais
• Palavras-chave em Português: Estatística; Análise de erros (Matemática); Estatística médica; Teoria bayesiana de decisão estatistica; Cromossomos humanos Anomalias; identificabilidade; erro de classificacao; estimador de Bayes; distribuicao Multinomial; Trissomia
• Link de acesso: https://hdl.handle.net/1843/RFFO-7KLRGF

The main causes of numerical chromosomal anomalies, including trisomies, arise from an error in the chromosomal segregation during the meiotic process, named a non-disjuntion. One of the most used techniques to analyse chromosomal anomalies is the Polymerase Chain Reaction (PCR) followed by a quantitative analysis via laser densitometry, which counts the number of peaks or alleles in polymorphic icrosatellite locus. It was shown in previous works that the number of peaks has a multinomial distribution whose parameters depend on the nondisjunction fraction '. In this work, we propose a misclassification model for estimating the meiosis I non-disjunction fraction '. We consider the Gauss Legendre method and de Simpson rule to extract information from the posterior distribution of '. Bayes estimators are comparedthrough Monte Carlo studies which focus in the influence of different sample sizes and differents probabilities of misclassification in the estimates. We apply the proposed method to estimate ' for patients with trisomy of chromosome 21 providing a sensitivity analysis for the method.In this case we use the Deviance Information Criterium (DIC) to compare the proposed model and the model proposed by Franco et al. (2003). The results obtained show that the proposed model is not the best. A possible reason for its low performance is the small proportion ofmisclassification.Key Words: Trisomy, Multinomial distribution, misclassification, identifiability, Bayes estimator.