A Classification-Based Machine Learning Approach for the Analysis of Genome-Wide Expression Data

Abstract

Three important areas of data analysis for global gene expression analysis are class discovery, class prediction, and finding dysregulated genes (biomarkers). The clinical application of microarray data will require marker genes whose expression patterns are sufficiently well understood to allow accurate predictions on disease subclass membership. Commonly used methods of analysis include hierarchical clustering algorithms, t-, F-, and Z-tests, and machine learning approaches. We describe an approach called the maximum difference subset (MDSS) algorithm that combines classification algorithms, classical statistics, and elements of machine learning and provides a coherent framework. By integrating prediction accuracy, the MDSS algorithm learns the critical threshold of statistical significance (the ␣ or P-value), eliminating the arbitrariness of setting a threshold of statistical significance and minimizing the effect of the normality assumptions. To reduce the false positive rate and to increase external validity of the predictive gene set, a jackknife step is used. This step identifies and removes genes in the initial MDSS with low combined predictive utility. The overall MDSS provides a prediction that is less dependent on an arbitrary study design (sample inclusion or exclusion) and should thus have high external validity. We demonstrate that this approach, unlike other published methods, identifies biomarkers capable of predicting the outcome of anthracycline-cytarabine chemotherapy in cases of acute myeloid leukemia. By incorporating two criteria-statistical significance and predictive utility-the approach learns the significance level relevant for a given data set. The MDSS approach can be used with any test and classifier operator pair.

Molecular Biomarkers for Quantitative and Discrete COPD Phenotypes

“…Our gene expression biomarkers were used to distinguish emphysema cases from controls in the cohort of Spira and colleagues by average linkage hierarchical clustering with Euclidean distance. These analyses were performed using the Gene Expression Data Analyzer (http://bioinformatics2.pitt.edu/ GE2/GEDA.html) (26). Cases were subjects who met the clinical criteria and underwent lung volume reduction surgery.…”

Identification of ATF-3, caveolin-1, DLC-1, and NM23-H2 as putative antitumorigenic, progesterone-regulated genes for ovarian cancer cells by gene profiling

“…We have used two methods Lyons-Weiler et al, 2003) to identify differentially expressed genes. Before data from the same slide or among different slides could be compared, it was necessary to normalize the data (i.e.…”

“…Figure 1 was generated by using a previously published analysis method . To further narrow down the gene list, we analysed data and identified dysregulated genes by using the maximum difference subset (MDSS) algorithm as described by Lyons-Weiler et al (2003). The advantages of this approach are (1) that it combines classification algorithms, classical statistics, and elements of machine learning, (2) it eliminates the arbitrariness of setting a threshold of statistical significance, and (3) by assimilating prediction accuracy, the MDSS algorithm is able to acquire the critical threshold of statistical significance (Pvalue).…”

Classification and Diagnostic Output Prediction of Cancer Using Gene Expression Profiling and Supervised Machine Learning Algorithms

“…Almost all genes among the selected 21 genes have been identified previously as containing abnormalities in AML or another form of leukemia. Most of the genes reported by Lyons-Weiler et al (2003) are also found in our DPLS gene set (HoxA9, PIG-B, MACH-alpha-2 protein, BPI Bactericidal/permeability increasing protein, Autoantigen PM-SCL, ERGIC-53 Protein, and so on). Figure 2(b) shows a heat map of the leukemia gene expression data based on the 21 selected genes most relevant for discrimination between success and failure of AML leukemia treatment.…”