非负矩阵分解在微阵列数据分类和聚类发现中的应用

摘要/Abstract

摘要：

基因芯片是微阵列技术的典型代表，它具有高通量的特性和同时检测全部基因组基因表达水平的能力。应用微阵列芯片的一个主要目的是基因表达模式的发现，即在基因组水平发现功能相似，生物学过程相关的基因簇；或者将样本分类，发现样本的各种亚型。例如根据基因表达水平对癌症样本进行分类，发现疾病的分子亚型。非负矩阵分解NMF方法是一种非监督的、非正交的、基于局部表示的矩阵分解方法。近年来这种方法被越来越多地应用在微阵列数据的分类分析和聚类发现中。系统地介绍了非负矩阵分解的原理、算法和应用，分解结果的生物学解释，分类结果的质量评估和基于NMF算法的分类软件。总结并评估了NMF方法在微阵列数据分类和聚类发现应用中的表现。

关键词: 非负矩阵分解, 微阵列数据, 分类分析, 聚类发现

Abstract:

A typical representation of microarray technologies is DNA microarray, which has ability to simultaneously measure the expression levels of all genes in genome due to its property of highthroughput. One of the main objectives of microarrays assay is gene expression pattern discovery, that is, not only the discovery of gene clusters where genes have similar functions or relative biological process, but also the discovery of sample subtypes which possess the intrinsic features, such as cancer subtypes. Nonnegative matrix factorization is an unsupervised, nonorthogonal, localbased representation methodology used into microarrays data analysis, especially in classification analysis and clustering discovery. The typical algorithm and some improved algorithms of NMF are introduced, and the biological annotation of factorization, the assessment of classification outcomes and the existing implementations basedon NMF are systematically summarized. Finally, the performance of NMF in recent microarray experiments is given.

Key words: non-negative matrix factorization;microarray data;classification analysis;clustering discovery

任重鲁，李金明. 非负矩阵分解在微阵列数据分类和聚类发现中的应用[J]. J4, 2014, 36(07): 1389-1397.

REN Zhonglu，LI Jinming. Application of nonnegative matrix factorization in
microarray data classification and clustering discovery [J]. J4, 2014, 36(07): 1389-1397.

参考文献

［1］Russell S, Meadows L, Russell R. Microarray technology in practice［M］.Xiao Huasheng,Zhang Chunxiu, Wu Xuemei, et al,translation. Beijing:Science Press,2010.(in Chinese)
［2］Huang Deshuang. Research on mining approaches for gene expression profiles data［M］.Beijing:Science Press,2009.(in Chinese)
［3］Jemal A, Bray F, Center M M, et al. Global cancer statistics［J］. CA Cancer J Clin, 2011, 61(2):6990.
［4］Valk P J M, Verhaak R G W, Beijen M A. Prognostically useful geneexpression profiles in acute myeloid leukemia［J］. The New England Journal of Medicine, 2004, 350:16171628.
［5］Barrier A, Boelle PY, Roser F, et al. Stage ii colon cancer prognosis prediction by tumor gene expression profiling［J］. Journal of Clinical Oncology, 2006, 24(29):46854691.
［6］Wang Y, Jatkoe T, Zhang Y, et al. Gene expression profiles and molecular markers to predict recurrence of dukes’b colon cancer［J］. Journal of Clinical Oncology, 2004, 22(9):15641571.
［7］The Cancer Genome Network. Integrated genomic analyses of ovarian carcinoma［J］. Nature, 2011, 474(7353):609615.
［8］Taylor B S,Schultz N,Hieronymus H,et al.Integrative genomic profiling of human prostate cancer［J］. Cancer Cell, 2010, 18(1):1122.
［9］Sorlie T, Perou C M, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications［J］. PNAS, 2001, 98(19):1086910874.
［10］Li A, Walling J, Ahn S, et al. Unsupervised analysis of transcriptomic profiles reveals six glioma subtypes［J］. Cancer Research, 2009, 69(5):20912099.
［11］Verhaak R G, Hoadley K A, Purdom E, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in pdgfra, idh1, egfr, and nf1［J］. Cancer Cell, 2010, 17(1):98110.
［12］Jorissen R N, Gibbs P, Christie M, et al. Metastasisassociated gene expression changes predict poor outcomes in patients with dukes stage b and c colorectal cancer［J］. Clinical Cancer Research, 2009, 15(24):76427651.
［13］Smith J J, Deane N G, Wu F, et al. Experimentally derived metastasis gene expression profile predicts recurrence and death in patients with colon cancer［J］. Gastroenterology, 2010, 138(3):958968.
［14］Vinciotti V, Tucker A, Kellam P, et al. Robust selection of predictive genes via a simple classifier［J］. Appl Bioinformatics, 2006, 5(1):111.
［15］Kossenkov A V, Ochs M F. Matrix factorization methods applied in microarray data analysis［J］. Data Mining and Bioinformatics, 2010, 4(1):7290.
［16］Lee D D, Seung H S. Learning the parts of objects by nonnegative matrix factorization［J］. Nature, 1999, 401:788791.
［17］Lee D D, Seung H S. Algorithms for nonnegative matrix factorization［J］. Adv. Neural Inform Process System, 2001, 13:556562.
［18］Devarajan K. Nonnegative matrix factorization:An analytical and interpretive tool in computational biology［J］. PLoS Computional Biology, 2008, 4(7):e1000029.
［19］Kim P M, Tidor B. Subsystem identification through dimensionality reduction of largescale gene expression data［J］. Genome Res, 2003, 13(7):17061718.
［20］Brunet J P, Tamayo P, Golub T R, et al. Metagenes and molecular pattern discovery using matrix factorization［J］. PNAS, 2004, 101(12):41644169.
［21］Frigyesi A, Hglund M. Nonnegative matrix factorization for the analysis of complex gene expression data, identification of clinically relevant tumor subtypes［J］. Cancer Informatics, 2008, 6:275292.
［22］Kim M H, Seo H J, Joung J G, et al. Comprehensive evaluation of matrix factorization methods for the analysis of DNA microarray gene expression data［J］. BMC Bioinformatics, 2011, 12(Suppl 13):S8.
［23］Gao Y, Church G. Improving molecular cancer class discovery through sparse nonnegative matrix factorization［J］. Bioinformatics, 2005, 21(21):39703975.
［24］Hoyer P O. Nonnegative sparse coding［C］∥Proc of Neural Networks for Signal Processing XII, 2002:557565.
［25］Hoyer P O. Nonnegative matrix factorization with sparseness constraints［J］. Journal of Machine Learning Research, 2004, 5:14571469.
［26］PascualMontano A,Carazo J M,Kochi K,et al.Nonsmooth nonnegative matrix factorization (nsnmf)［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(3):403415.
［27］Liu W, Zheng N, Lu X. Nonnegative matrix factorization for visual coding［C］∥Proc of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2003:293296.
［28］CarmonaSaez P, PascualMarqui R D, Tirado F, et al. Biclustering of gene expression data by nonsmooth nonnegative matrix factorization［J］. BMC Bioinformatics, 2006, 7:7896.
［29］MejiaRoa E, CarmonaSaez P, Nogales R, et al. Bionmf:A webbased tool for nonnegative matrix factorization in biology［J］. Nucleic Acids Res, 2008, 36(Web Server issue):W523W528.
［30］PascualMontano A, CarmonaSaez P, Chagoyen M, et al. Bionmf:A versatile tool for nonnegative matrix factorization in biology［J］. BMC Bioinformatics, 2006, 7:366374.
［31］Li Y, Sima D M, Cauter S V, et al. Hierarchical nonnegative matrix factorization (hnmf):A tissue pattern differentiation method for glioblastoma multiforme diagnosis using mrsi［J］. NMR Biomed, 2013, 26(3):307319.
［32］OrtegaMartorell S, Lisboa P J, Vellido A, et al. Convex nonnegative matrix factorization for brain tumor delimitation from mrsi data［J］. PLoS One, 2012, 7(10):e47824.
［33］Lee C M, Mudaliar M A, Haggart D, et al. Simultaneous nonnegative matrix factorization for multiple large scale gene expression datasets in toxicology［J］. PLoS One, 2012, 7(12):e48238.
［34］Shi Jinlong, Luo Zhigang. Research on the advances of nonnegative matrix factorization and its application in bioinformatics［J］. Computer Engineering & Science, 2010, 32(8):117123.(in Chinese)
［35］Monti S, Tamayo P, Mesirov J, et al. Consensus clustering:A resamplingbased method for class discovery and visualization of gene expression microarray data［J］. Machine Learning, 2003, 52:91118.
［36］Hutchins L N, Murphy S M, Singh P, et al. Positiondependent motif characterization using nonnegative matrix factorization［J］. Bioinformatics, 2008, 24(23):26842690.
［37］Tibshirani R, Walther G, Hastie T. Estimating the number of clusters in a data set via the gap statistic［J］. J.R.Statist, 2001, 63:411423.
［38］Rousseeuw P. Silhouettes:A graphical aid to the interpretation and validation of cluster analysis［J］. Journal of Computational and Applied Mathematics, 1987, 20:5365.
［39］Frades I, Matthiesen R. Overview on techniques in cluster analysis［J］. Methods in Molecular Biology, 2010, 593:81107.
［40］Kim H, Park H. Sparse nonnegative matrix factorizations via alternating nonnegativityconstrained least squares for microarray data analysis［J］. Bioinformatics, 2007, 23(12):14951502.
［41］Wang G, Kossenkov A V, Ochs M F. Lsnmf:A modified nonnegative matrix factorization algorithm utilizing uncertainty estimates［J］. BMC Bioinformatics, 2006, 7:175184.
［42］Reich M, Liefeld T, Gould J, et al. Genepattern 2.0［J］. Nature Genetics, 2006, 38:500501.
［43］Qi Q, Zhao Y, Li M, et al. Nonnegative matrix factorization of gene expression profiles:A plugin for brbarraytools［J］. Bioinformatics, 2009, 25(4):545547.
［44］Gaujoux R, Seoighe C. A flexible r package for nonnegative matrix factorization［J］. BMC Bioinformatics, 2010, 11:367375.
［45］Gentleman R C, Carey V J, Bates D M, et al. Bioconductor:Open software development for computational biology and bioinformatics［J］. Genome Biology, 2004, 5(10):R80.81R80.16.
［46］Slonim D K, Tamayo P, Mesirov J P, et al. Class prediction and discovery using gene expression data［C］∥Proc of the 4th International Conference on Computational Molecualr Biology, 2000:236272.
［47］Pomeroy S L, Tamayo P, Gaasenbeek M, et al. Prediction of central nervous system embryonal tumour outcome based on gene expression［J］. Nature, 2002, 415:436442.
［48］Wang Y, Jia Y, Hu C, et al. Fisher nonnegative matrix factorization for learning local features［C］∥Proc of Asian Conference on Computer Vision, 2004:2730.
［49］Tibshirani R, Hastie T, Narasimhan B, et al. Diagnosis of multiple cancer types by shrunken centroids of gene expression［J］. PNAS, 2002, 99(10):65676572.
［50］Dabney A R. Clanc:Pointandclick software for classifying microarrays to nearest centroids［J］. Bioinformatics, 2006, 22(1):122123.
［51］Networks C G A. Comprehensive molecular characterization of human colon and rectal cancer［J］. Nature, 2012, 487(7407):330337.
［52］Govi S, Dognini G P, Licata G, et al. Nonnegative matrix factorization to perform unsupervised clustering of genome wide DNA profiles in mature b cell lymphoid neoplasms［J］. Br J Haematol, 2010, 150(2):226229.
［53］Inamura K, Fujiwara T, Hoshida Y, et al. Two subclasses of lung squamous cell carcinoma with different gene expression profiles and prognosis identified by hierarchical clustering and nonnegative matrix factorization［J］. Oncogene, 2005, 24(47):71057113.
［54］Jiang X, Langille M G, Neches R Y, et al. Functional biogeography of ocean microbes revealed through nonnegative matrix factorization［J］. PLoS One, 2012, 7(9):e43866.
［55］Wang H M, Hsiao C L, Hsieh A R, et al. Constructing endophenotypes of complex diseases using nonnegative matrix factorization and adjusted rand index［J］. PLoS One, 2012, 7(7):e40996.
附中文参考文献：
［1］史蒂夫·拉塞尔，莉萨·梅多斯，罗斯林·拉塞尔. 生物芯片技术与实践(中文版)［M］. 肖华胜，张春秀，武雪梅，等译.北京:科学出版社,2010.
［2］黄德双. 基因表达谱数据挖掘方法研究［M］. 北京:科学出版社,2009.
［34］石金龙, 骆志刚. 非负矩阵算法及其在生物信息学中的应用［J］. 计算机工程与科学,2010,32（8）:117123.

[1]	贾俊杰, 姚叶旺, 陈旺虎. 基于非负矩阵分解的群组推荐算法[J]. 计算机工程与科学, 2022, 44(05): 933-943.
[2]	曹文态, 杨德刚, 冯骥, . 基于扩展自然邻居的无参分类方法[J]. 计算机工程与科学, 2021, 43(09): 1692-1699.
[3]	孙刚1,2，张靖1,3. 面向高维微阵列数据的集成特征选择算法[J]. J4, 2016, 38(07): 1330-1337.
[4]	石金龙,骆志刚. 非负矩阵分解算法及其在生物信息学中的应用研究[J]. J4, 2010, 32(8): 117-123.
[5]	晁永国. 一种改进的非负稀疏编码图像编码方案[J]. J4, 2010, 32(10): 66-68.
[6]	韩宗芬余潮孙建华杨艳屏. 一种基于组件结构的协同检测模型[J]. J4, 2005, 27(7): 10-12.
[7]	刘青[1] 李海峰[2] 袁科[3]. 微阵列数据中一种改进Bagging决策树算法的研究[J]. J4, 2005, 27(6): 78-80.