论文部分内容阅读
由于基因间的调控和相互作用表现为功能基因组合的形式,在对样本的分类能力是以特征集合的形式整体体现出来的.由此,考察由多个基因构成的基因簇作为区分常人和癌症患者的分类因素,利用独立成分分析(ICA)技术最大程度地降低基因之间的相互影响,从而获得基因簇信息.随后采用了支持向量机,依据提取出的基因簇进行分类,筛选出致病的癌症基因.为了能够得到最好的分类因素,将问题转化为稀疏表示的优化问题.此外,还利用含噪声的ICA和带松弛因子的非光滑优化模型来研究含噪声的基因图谱.最后,借助于条件概率模型,将临床结论与基因图谱相结合,对病人数据进行了筛选.
Because the regulation and interaction between genes appear as the form of functional genomics, the ability to classify the samples is embodied as a collection of features, thus examining the gene cluster formed by multiple genes as a marker for distinguishing common people from cancers (ICA) technique to minimize the interaction between genes and get the information of gene cluster.Then the support vector machine (SVM) was used to classify the samples according to the extracted gene cluster to screen out the pathogenic In order to get the best classification factor, the problem is transformed into the problem of sparse representation optimization.In addition, the noisy ICA and the non-smooth optimization model with relaxation factor are also used to study the noise-containing gene map.Finally, With the help of the conditional probability model, the clinical findings were combined with the genetic map to screen patient data.