Abstract:Integrated analysis of human gene expression data from multiple studies has become essential in genomics research for complex traits. However, integrating data generated from different cohorts with different platforms, such as microarray and RNA-seq, often requires data preprocessing, including normalization. In this study, we empirically evaluate 9 commonly used cross-platform normalization methods. We classify these methods into two main types: joint and separate normalization. Joint methods normalize multiple datasets together, while separate methods normalize each dataset independently. We further divide these methods into unsupervised and supervised approaches depending on whether they use outcomes during their normalization process. Examples of joint unsupervised methods include Quantile Normalization (QN), while Rank-in serves as an example of a joint supervised method. Training Distribution Matching (TDM) is an example of a separate unsupervised method. We assess each method's ability to cluster samples, predict outcomes, and detect differentially expressed (DE) genes using three real datasets and simulated data. First, our real data analysis suggests that while joint supervised methods can cluster sample groups better than the other two method groups, they double use the outcome data with artificially inflated clustering performance. Their biases are further demonstrated by their inflated type I error in DE analysis and clustering results from simulated data with no DE genes. Second, for outcome prediction, supervised normalization is no longer applicable. Among the unsupervised methods, QN significantly outperforms the other approaches, regardless of whether RNA-seq data is used to predict microarray outcomes or vice versa. Finally, we compare normalization methods on downstream DE analysis using simulation. In addition to direct DE analysis on the combined normalized data using the non-parametric Wilcoxon rank-sum test, we also perform a meta-analysis that combines p-values of the DE analysis from each individual data. For DE analysis, the meta-analysis consistently achieves the best balance between controlling type I error and maximizing power for DE gene detection. Our research suggests that while normalization is critical for the integrated analysis of transcriptomics data, simple QN is the most efficient and unbiased normalization approach for outcome prediction, and meta-analysis is the most appropriate for DE analysis.

Extensive increase of microarray signals in cancers calls for novel normalization assumptions

Extensive Up-Regulation of Gene Expression in Cancer: the Normalised Use of Microarray Data

Revealing Extensive Up - Regulation of Gene Expression in Pancreatic Cancer Based on Microarray Data

A Novel Approach to Detect Differentially Expressed Genes from Count-Based Digital Databases by Normalizing with Housekeeping Genes.

Comparison of Different Normalization Assumptions for Analyses of Dna Methylation Data from the Cancer Genome

Analyses of DNA Methylation Data Based on Different Normalization Assumptions

A Normalization-Free and Nonparametric Method Sharpens Large-Scale Transcriptome Analysis and Reveals Common Gene Alteration Patterns in Cancers

Use of normalization methods for analysis of microarrays containing a high degree of gene effects

Ratio Adjustment and Calibration Scheme for Gene-Wise Normalization to Enhance Microarray Inter-Study Prediction

Common Human Cancer Genes Discovered by Integrated Gene-Expression Analysis

cDNA Microarray Experiment: Design Issues in Early Stage and the Need of Normalization

Evaluating Cross-Platform Normalization Methods for Integrated Microarray and RNA-seq Data Analysis

Normalization Benefits Microarray-Based Classification

Deciphering Global Signal Features of High-Throughput Array Data from Cancers

The effect of GeneChip gene definitions on the microarray study of cancers.

Molecular Signature of Cancer at Gene Level or Pathway Level? Case Studies of Colorectal Cancer and Prostate Cancer Microarray Data.

Modified least-variant set normalization for miRNA microarray.

Apparently low reproducibility of true differential expression discoveries in microarray studies.

Different Normalization Strategies for Microarray Gene Expression Traits Affect the Heritability Estimation

An Improved Normalized Signal to Noise Ratio Method for Irrelevant Genes Removing

Global Differential Gene Expression in Cancers and its Implications for Building Robust Diagnostic Classifiers