A compressed sensing based approach for subtyping of leukemia from gene expression data.
A compressed sensing based approach for subtyping of leukemia from gene expression data.
复制标题
DOI：
10.1142/s0219720011005689
复制
发表时间：
2011-10
影响因子：
1
通讯作者：
Wang YP
中科院分区：
文献类型：
--
作者：
Tang W;Cao H;Duan J;Wang YP
随着基因组技术的发展，对能够有效处理高通量全基因组数据的新方法的需求比以往任何时候都要强烈。压缩感知是统计学和信号处理领域的一种新兴方法。利用CS理论，可以从信号的稀疏表示中唯一地重构或近似信号，从而可以更好地区分不同类型的信号。然而，CS方法在全基因组数据分析中的应用却很少被研究。我们提出了一种新的基于CS的基因组数据分类方法，并通过基因表达分析测试其在白血病亚型中的性能。根据不同的遗传标记检测癌症的亚型，如白血病是重要的，这为治疗的个体化和治疗的改进带来了希望。在我们的工作中，四个统计特征被用来选择显著的基因进行分类。我们从7，129个基因中选择了这些基因，所提出的CS方法在交叉验证评估时达到了97.4%的分类准确率，在另一个独立数据集评估时达到了94.3%。该方法对噪声的鲁棒性也进行了测试，给出了良好的性能。因此，这项工作表明CS方法可以有效地检测白血病的亚型，这意味着提高了白血病诊断的准确性。
With the development of genomic techniques, the demand for new methods that can handle high-throughput genome-wide data effectively is becoming stronger than ever before. Compressed sensing (CS) is an emerging approach in statistics and signal processing. With the CS theory, a signal can be uniquely reconstructed or approximated from its sparse representations, which can therefore better distinguish different types of signals. However, the application of CS approach to genome-wide data analysis has been rarely investigated. We propose a novel CS-based approach for genomic data classification and test its performance in the subtyping of leukemia through gene expression analysis. The detection of subtypes of cancers such as leukemia according to different genetic markups is significant, which holds promise for the individualization of therapies and improvement of treatments. In our work, four statistical features were employed to select significant genes for the classification. With our selected genes out of 7,129 ones, the proposed CS method achieved a classification accuracy of 97.4% when evaluated with the cross validation and 94.3% when evaluated with another independent data set. The robustness of the method to noise was also tested, giving good performance. Therefore, this work demonstrates that the CS method can effectively detect subtypes of leukemia, implying improved accuracy of diagnosis of leukemia.
登录
查看更多内容
影响因子：
56.9
作者：
Golub, TR;Slonim, DK;Lander, ES
通讯作者：
Lander, ES
影响因子：
20.3
作者：
Starczynowski, Daniel T.;Vercauteren, Suzanne;Karsan, Aly
通讯作者：
Karsan, Aly
DOI：
10.1155/2009/162824
发表时间：
2009
期刊：
EURASIP journal on bioinformatics & systems biology
影响因子：
--
作者：
Dai W;Sheikh MA;Milenkovic O;Baraniuk RG
通讯作者：
Baraniuk RG
影响因子：
14.9
作者：
Baraniuk, Richard G.
通讯作者：
Baraniuk, Richard G.
影响因子：
6.5
作者：
van der Gaag, EJ;Leccia, MT;Byers, HR
通讯作者：
Byers, HR