Download full-text

Full-text

Available from: Robert Clarke, Jun 14, 2014
0 Followers
 · 
152 Views
  • Source
    • "Such genes may compromise multicategory classification accuracy, especially when a small gene subset is chosen. It is also important to note that, while univariate or multivariate analysis methods using complex criterion functions may reveal subtle marker effects (Cai et al., 2007; Liu et al., 2005; Xuan et al., 2007; Zhou and Tuck, 2007), they are also prone to overfitting. Recent studies have found that for small sample sizes, univariate methods fared comparably to multivariate methods (Lai et al., 2006; Shedden et al., 2003) and simple fold change analysis produced more reproducible marker genes than significance analysis of variance-incorporated t-tests (Shi et al., 2008). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Microarray gene expressions provide new opportunities for molecular classification of heterogeneous diseases. Although various reported classification schemes show impressive performance, most existing gene selection methods are suboptimal and are not well-matched to the unique characteristics of the multicategory classification problem. Matched design of the gene selection method and a committee classifier is needed for identifying a small set of gene markers that achieve accurate multicategory classification while being both statistically reproducible and biologically plausible. We report a simpler and yet more accurate strategy than previous works for multicategory classification of heterogeneous diseases. Our method selects the union of one-versus-everyone (OVE) phenotypic up-regulated genes (PUGs) and matches this gene selection with a one-versus-rest support vector machine (OVRSVM). Our approach provides even-handed gene resources for discriminating both neighboring and well-separated classes. Consistent with the OVRSVM structure, we evaluated the fold changes of OVE gene expressions and found that only a small number of high-ranked genes were required to achieve superior accuracy for multicategory classification. We tested the proposed PUG-OVRSVM method on six real microarray gene expression data sets (five public benchmarks and one in-house data set) and two simulation data sets, observing significantly improved performance with lower error rates, fewer marker genes, and higher performance sustainability, as compared to several widely-adopted gene selection and classification methods. The MATLAB toolbox, experiment data and supplement files are available at http://www.cbil.ece.vt.edu/software.htm.
    Journal of Machine Learning Research 08/2010; 11:2141-2167. · 2.85 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Normalization is a prerequisite for almost all follow-up steps in microarray data analysis. Accurate normaliza-tion across different experiments and phenotypes assures a common base for comparative yet quantitative studies using gene expression data. In this paper, we report a novel normalization approach, namely iterative nonlinear regression (INR) method, which exploits concurrent identification of invariantly expressed genes (IEGs) and implementation of nonlinear regression normalization. The INR scheme features an iterative process that performs the following two steps alterna-tively: (1) selection of IEGs and (2) estimation of nonlinear regression function for normalization. We demonstrate the principle and performance of the INR approach on two real microarray data sets. As compared to major peer methods (e.g., linear regression method, Loess method and iterative ranking method), INR method shows an improved perform-ance in achieving low expression variance across replicates and excellent fold-change preservation for differently ex-pressed genes.
    The Open Applied Informatics Journal 11/2007; 107(11):11-19. DOI:10.2174/1874136300701010011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The main limitations of most existing clustering methods used in genomic data analysis include heuristic or random algorithm initialization, the potential of finding poor local optima, the lack of cluster number detection, an inability to incorporate prior/expert knowledge, black-box and non-adaptive designs, in addition to the curse of dimensionality and the discernment of uninformative, uninteresting cluster structure associated with confounding variables. In an effort to partially address these limitations, we develop the VIsual Statistical Data Analyzer (VISDA) for cluster modeling, visualization, and discovery in genomic data. VISDA performs progressive, coarse-to-fine (divisive) hierarchical clustering and visualization, supported by hierarchical mixture modeling, supervised/unsupervised informative gene selection, supervised/unsupervised data visualization, and user/prior knowledge guidance, to discover hidden clusters within complex, high-dimensional genomic data. The hierarchical visualization and clustering scheme of VISDA uses multiple local visualization subspaces (one at each node of the hierarchy) and consequent subspace data modeling to reveal both global and local cluster structures in a "divide and conquer" scenario. Multiple projection methods, each sensitive to a distinct type of clustering tendency, are used for data visualization, which increases the likelihood that cluster structures of interest are revealed. Initialization of the full dimensional model is based on first learning models with user/prior knowledge guidance on data projected into the low-dimensional visualization spaces. Model order selection for the high dimensional data is accomplished by Bayesian theoretic criteria and user justification applied via the hierarchy of low-dimensional visualization subspaces. Based on its complementary building blocks and flexible functionality, VISDA is generally applicable for gene clustering, sample clustering, and phenotype clustering (wherein phenotype labels for samples are known), albeit with minor algorithm modifications customized to each of these tasks. VISDA achieved robust and superior clustering accuracy, compared with several benchmark clustering schemes. The model order selection scheme in VISDA was shown to be effective for high dimensional genomic data clustering. On muscular dystrophy data and muscle regeneration data, VISDA identified biologically relevant co-expressed gene clusters. VISDA also captured the pathological relationships among different phenotypes revealed at the molecular level, through phenotype clustering on muscular dystrophy data and multi-category cancer data.
    BMC Bioinformatics 10/2008; 9(1):383. DOI:10.1186/1471-2105-9-383 · 2.67 Impact Factor
Show more