Publicação
Microarray gene expression data integration: an application to brain tumor grade determination
| Resumo: | World Health Organization ranks brain tumors in four stages, being the fourth grade the most aggressive. Glioblastoma, a fourth grade tumor, is one of the most severe human diseases that almost inevitability leads to death. Physicians address the classification in grades through direct inspection. Indeed, there is a need for good automatic predictors of tumor grade, which are not affected by human misclassification errors and that can be made with less invasive diagnostic tools. This work address the stages involved in the process of selecting a good tumor grade predictor, based on microarray gene expression data. In this work, the information integration from heterogeneous platforms is highlighted, evidencing the particularities of choosing approaches working at gene, transcript or probeset levels. Distinct machine learning algorithms and integration methods are tested, analyzing their ability to produce a good set of predictors for tumor grade. |
|---|---|
| Autores principais: | Valente, Eduardo |
| Outros Autores: | Rocha, Miguel |
| Assunto: | Gene expression Glioblastoma Microarrays data integration Filter methods Wrapper methods |
| Ano: | 2015 |
| País: | Portugal |
| Tipo de documento: | comunicação em conferência |
| Tipo de acesso: | acesso restrito |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | World Health Organization ranks brain tumors in four stages, being the fourth grade the most aggressive. Glioblastoma, a fourth grade tumor, is one of the most severe human diseases that almost inevitability leads to death. Physicians address the classification in grades through direct inspection. Indeed, there is a need for good automatic predictors of tumor grade, which are not affected by human misclassification errors and that can be made with less invasive diagnostic tools. This work address the stages involved in the process of selecting a good tumor grade predictor, based on microarray gene expression data. In this work, the information integration from heterogeneous platforms is highlighted, evidencing the particularities of choosing approaches working at gene, transcript or probeset levels. Distinct machine learning algorithms and integration methods are tested, analyzing their ability to produce a good set of predictors for tumor grade. |
|---|