23 October 2007 in GeoInformatica
1 January 2005 in BioMed Research International
Gene selection is an important issue in analyzing multiclass microarray data. Among many proposed selection methods, the traditional ANOVA F test statistic has been employed to identify informative genes for both class prediction (classification) and discovery problems. However, the F test statistic assumes an equal variance. This assumption may not be realistic for gene expression data. This paper explores other alternative test statistics which can handle heterogeneity of the variances. We study five such test statistics, which include Brown‐Forsythe test statistic and Welch test statistic. Their performance is evaluated and compared with that of F statistic over different classification methods applied to publicly available microarray datasets.
1 January 2009 in BioMed Research International
Accurate prediction of survival rates of cancer patients is often key to stratify patients for prognosis and treatment. Survival prediction is often accomplished by the TNM system that involves only three factors: tumor extent, lymph node involvement, and metastasis. This prediction from the TNM has been limited, because other potential prognostic factors are not used in the system. Based on availability of large cancer datasets, it is possible to establish powerful prediction systems by using machine learning procedures and statistical methods. In this paper, we present an ensemble clustering‐based approach to develop prognostic systems of cancer patients. Our method starts with grouping combinations that are formed using levels of factors recorded in the data. The dissimilarity measure between combinations is obtained through a sequence of data partitions produced by multiple use of PAM algorithm. This dissimilarity measure is then used with a hierarchical clustering method in order to find clusters of combinations. Prediction of survival is made simply by using the survival function derived from each cluster. Our approach admits multiple factors and provides a practical and useful tool in outcome prediction of cancer patients. A demonstration of use of the proposed method is given for lung cancer patients.
1 January 2007 in BMC Genomics
Gene expression profiles based on microarray data have been suggested by many studies as potential molecular prognostic indexes of breast cancer. However, due to the confounding effect of clinical background, independent studies often obtained inconsistent results. The current study investigated the possibility to improve the quality and generality of expression profiles by integrated analysis of multiple datasets. Profiles of recurrence outcome were derived from two independent datasets and validated by a third dataset.
21 December 2010 in BMC Bioinformatics
Most genomic data have ultra-high dimensions with more than 10,000 genes (probes). Regularization methods with L1 and Lp penalty have been extensively studied in survival analysis with high-dimensional genomic data. However, when the sample size n ≪ m (the number of genes), directly identifying a small subset of genes from ultra-high (m > 10, 000) dimensional data is time-consuming and not computationally efficient. In current microarray analysis, what people really do is select a couple of thousands (or hundreds) of genes using univariate analysis or statistical tests, and then apply the LASSO-type penalty to further reduce the number of disease associated genes. This two-step procedure may introduce bias and inaccuracy and lead us to miss biologically important genes.