Prediction by using selected top genes
| Autor: | Yu-Nong Lin, 林雨農 |
|---|---|
| Rok vydání: | 2009 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 97 DNA microarray is a useful and important tool in biology research, especially in cancer research. Except for survival data, we can also use the gene expression data from DNA microarray to identify the possible factors associated with the cancer. In gene expression data, there are large number of genes and small number of samples. Therefore, when we construct a prediction model and estimate the parameters, we will face with difficulty in computation. In order to solve this problem, we evaluated the effect of the number of top ranked genes. We used two gene ranking methods, the statistics p-values and Cox scores, and used three dimension reduction method, the principal component analysis, the supervised principal component analysis and the partial least squares approach, to build the prediction model with the top ranked gene of training data, then we used the same top genes to compare the results of training data with test data. Lastly, we compared the performance of the above methods. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|