Meta-analysis approach as a gene selection method in class prediction: does it improve model performance? A case study in acute myeloid leukemia

Abstract

BackgroundAggregating gene expression data across experiments via meta-analysis is expected to increase the precision of the effect estimates and to increase the statistical power to detect a certain fold change. This study evaluates the potential benefit of using a meta-analysis approach as a gene selection method prior to predictive modeling in gene expression data.ResultsSix raw datasets from different gene expression experiments in acute myeloid leukemia (AML) and 11 different classification methods were used to build classification models to classify samples as either AML or healthy control. First, the classification models were trained on gene expression data from single experiments using conventional supervised variable selection and externally validated with the other five gene expression datasets (referred to as the individual-classification approach). Next, gene selection was performed through meta-analysis on four datasets, and predictive models were trained with the selected genes on the fifth dataset and validated on the sixth dataset. For some datasets, gene selection through meta-analysis helped classification models to achieve higher performance as compared to predictive modeling based on a single dataset; but for others, there was no major improvement. Synthetic datasets were generated from nine simulation scenarios. The effect of sample size, fold change and pairwise correlation between differentially expressed (DE) genes on the difference between MA- and individual-classification model was evaluated. The fold change and pairwise correlation significantly contributed to the difference in performance between the two methods. The gene selection via meta-analysis approach was more effective when it was conducted using a set of data with low fold change and high pairwise correlation on the DE genes.ConclusionGene selection through meta-analysis on previously published studies potentially improves the performance of a predictive model on a given gene expression data.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-017-1619-7) contains supplementary material, which is available to authorized users.

High-Throughput Omics and Statistical Learning Integration for the Discovery and Validation of Novel Diagnostic Signatures in Colorectal Cancer

“…Several systematic approaches and guidelines have been proposed for selecting important biomarker candidates across experiments [12]. In the circumstance that a single gene expression analysis might not provide a reliable and generalizable conclusion, the quantitative analysis of the combined datasets from multiple sources and technologies appears to be an efficient solution to increase the sample size and enhance the statistical power and thus possibly helps identify more clinically relevant biomarkers [13,14]. Choosing sufficient control samples and appropriate data mining techniques, removing batch effects across different platforms and studies are required to ensure a robust process of biomarker discovery and validation [15].…”

A Novel Efficient Feature Dimensionality Reduction Method and Its Application in Engineering

“…For instance, leukemia has 7129 gene dimensions but only 72 samples for analysis, which increases the challenges of feature dimensionality reduction. Various methods exist for gene selection, but they are computationally costly and complex [15][16][17][18]. Many researchers have introduced information to facilitate gene selection, but these methods cannot properly reflect the length of a set of gene features, namely, the number of dimensions after dimensionality reduction.…”

The importance of genomic predictors for clinical outcome of hematological malignancies

“…Supervised learning is used to identify genes related to known categories such as cancer type or clinical outcome, and unsupervised learning is used to explore the similarity of gene expression patterns 6 . A large number of supervised learning have been used to explore hematological malignancies, including weighted voting, k-nearest neighbors, support vector machines, artificial neural networks, decision trees, random forest, and nearest shrunken centroid algorithms 7–10 . For unsupervised learning, K-means clustering, principal component analysis, nonnegative matrix factorization, and weighted co-expression network analysis (WGCNA) have been widely used to investigate hematological malignancies 11–14 .…”

“…6 A large number of supervised learning have been used to explore hematological malignancies, including weighted voting, k-nearest neighbors, support vector machines, artificial neural networks, decision trees, random forest, and nearest shrunken centroid algorithms. [7][8][9][10] For unsupervised learning, Kmeans clustering, principal component analysis, nonnegative matrix factorization, and weighted co-expression network analysis (WGCNA) have been widely used to investigate hematological malignancies. [11][12][13][14] However, when conducting biological exploration, no formal classifier is needed to observe the correlation between 2 genes and investigate 1 gene's effect on the prognosis of patients.…”