摘要
针对癌症样本类型存在的局部相关性,提出了一种新颖的基因特征抽取算法。利用基因表达谱中基因表达数据的空间映射,提取不同类型癌症的致癌因子,并构建癌症组的关联空间,在关联空间之上抽取分类基因特征。理论证明可以有效识别癌症模式,并在急性白血病数据集上进行实验,结果表明利用抽取的分类基因特征可以有效识别AML/ALL。
Using the cancerogenic factor's local correlation to a type of cancer,a novel method of gene features extraction is proposed.Using gene features transformation,the cancerogenic factors are extracted to different cancers and a relative space is built to the cancer,then the gene features are extracted for cancers with them.The cancer pattern is proved to can be recognized.The experiment is explored in the Leukemia dataset and Colon dataset,the results show that AML/ALL and to Tumor Colon Tissue(TCT)/normal Colon Tissue(NCT) are recognized effectively.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第30期237-240,244,共5页
Computer Engineering and Applications
基金
国家自然科学基金No.60873184
国家博士后科学基金No.20100471790
湖南省自然科学基金No.07JJ5085~~
关键词
基因表达谱
基因特征抽取
癌症分类
gene expression profile; gene feature extraction; cancer recognition;