RegiÃes gÃnicas de interesse para o desenvolvimento de um diagnÃstico do Papillomavirus Humano: abordagem in silico

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

Cervical cancer is the second most common cancer in women and the seventh in the world with 493,000 of new cases per year, and mortality between 50 and 55%. In Brazil, the estimate is 19,260 new cases for 2006, with an estimated risk for 20.31 cases to each 100,000 women. In Brazil, as well as in the others developing countries, the frequency of the Pap smear are low and the cases normally are diagnosed in advanced phase. In the whole world the Human Papillomavirus (HPV), associated with cervical cancer, is a public health problem. The genome analysis, through comparative genomics, of the high and low risk HPVs, can indicate important regions associated in carcinogenicity. This study proposes a methodology to classify the carcinogenicity risk of HPVs by using protein sequences of five genes, E6, E1, E7, L1, and L2, which are present in 78 HPV genomes available in the public databases. Using Hidden Markov Models (HMM) twenty-four regions had been selected in the five proteins. Eight, of 24 regions, were selected when used a discriminative model based on Support Vector Machine (SVM) coupled with a Genetic Algorithm (GA). Different combinations of these eight regions produced a set of classifiers capable of correctly predicting 100% of HPV risk type. This technique allowed determining regions that can have an important role in the carcinogenicity of the HPV. These results suggest regions of interest for developing a method of diagnosis for viral DNA presence, contributing in the prevention of cervical cancer

ASSUNTO(S)

bioinformÃtica ciencias biologicas papillomavirus bioinformatics in silico papillomavirus in silico molecular diagnostics diagnÃstico molecular

ACESSO AO ARTIGO

Documentos Relacionados