Gene selection and classification of microarray data using random forest

By Ramón Díaz-Uriarte, S. Á. D. Andrés. 2006

2,984 citations recorded by the source corpus.

BackgroundSelection of relevant genes for sample classification is a common task in most gene expression studies, where researchers try to identify the smallest possible set of genes that can still achieve good predictive performance (for instance, for future use with diagnostic purposes in clinical practice). Many gene selection approaches use univariate (gene-by-gene) rankings of gene relevance and arbitrary thresholds to select the number of genes, can only be applied to two-class problems, and use gene selection ranking criteria unrelated to the classification algorithm. In contrast, random forest is a classification algorithm well suited for microarray data: it shows excellent performance even when most predictive variables are noise, can be used when the number of variables is much larger than the number of observations and in problems involving more than two classes, and returns measures of variable importance. Thus, it is important to understand the performance of random forest with microarray data and its possible use for gene selection.ResultsWe investigate the use of random forest for classification of microarray data (including multi-class problems) and propose a new method of gene selection in classification problems based on random forest. Using simulated and nine microarray data sets we show that random forest has comparable performance to other classification methods, including DLDA, KNN, and SVM, and that the new gene selection procedure yields very small sets of genes (often smaller than alternative methods) while preserving predictive accuracy.ConclusionBecause of its performance and features, random forest and gene selection using random forest should probably become part of the "standard tool-box" of methods for class prediction and gene selection with microarray data.

Subfields: Information Systems, Molecular Biology, Computational Biology & Bioinformatics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Systems

In Molecular Biology

In Computational Biology & Bioinformatics

Citations in the corpus

Consistent Individualized Feature Attribution for Tree Ensembles (2018, methodology): “Different choices about the method of feature value permutation lead to variations of this basic approach [1, 10, 14, 23, 26].”
Uniform Convergence of Random Forests via Adaptive Concentration (2015, background): “Trees [10] and random forests [8] are among the most widely used machine learning predictors today, with applications in a broad variety of fields such as ecology [14, 33], genetics [17, 39], and remote sensing [20, 32].”
A balanced iterative random forest for gene selection from microarray data (2013, methodology): “Finally, and as in [7], the last issue addressed in this paper is the evaluation of the selected genes to determine whether they are stable and appear in multiple executions, or selected only once.”
Zeileis Conditional Variable Importance for Random Forests (2015, methodology): “(page number not for citation purposes) since recent results [16,17] indicate that the raw importance VI(Xj) has better statistical properties, we will only consider the unscaled version here.”
Empirical characterization of random forest variable importance measures (2008, methodology): “Further, Díaz-Uriarte and AlvarezdeAndrés (2006) reported the error estimates of the RF algorithms using their gene selection method, but they did not report variable importance measures for the known true predictors in the simulated data sets.”
A Brief Review of Random Forests for Water Scientists and Practitioners and Their Recent History in Water Resources (2019, methodology): “4, and Biau and Scornet [2], Díaz-Uriarte and De Andres [80]).”
Boruta – A System for Feature Selection (2010, methodology): “The relative performance of SVM and RF in various bioinformatical applications is disputed [6, 15], nevertheless the RF method provides two key advantages: it does not need any parameter optimisation to reach its highest performance and it has the built-in statistical estimation of the variable impo”
Functional Knowledge Transfer for High-accuracy Prediction of Under-studied Biological Processes (2013, background): “In genomic applications, Random forest has gained interest due to the many high-dimensional genomic learning problems [93].”
Correlation and variable importance in random forests (2013, methodology): “For instance, Díaz-Uriarte and Alvarez de Andrés [10] and Genuer et al. [11] use the out-of-bag (OOB) error estimate whereas Jiang et al. [18] use both the OOB and a validation set to estimate the error.”