Abstract
Faced with the lack of reliability and reproducibility in omics studies, more careful and robust methods are needed to overcome the existing challenges in the multi-omics analysis. In conventional omics data analysis, signal intensity values (denoted by M and values) are estimated neglecting pixel-level uncertainties, which may reflect noise and systematic artifacts. For example, intensity values from two-color microarray data are estimated by taking the mean or median of the pixel intensities within the spot and then subjected to a within-slide normalization by LOWESS. Thus, focusing on estimation and normalization of gene expression profiles, we propose a spot quantification method that takes into account pixel-level variability. Also, to preserve relevant variation that may be removed in LOWESS normalization with poorly chosen parameters, we propose a parameter selection method that is parsimonious and considers intrinsic characteristics of microarray data, such as heteroskedasticity. The usefulness of the proposed methods is illustrated by an application to real intestinal metaplasia data. Compared with the conventional approaches, the analysis is more robust and conservative, identifying fewer but more reliable differentially expressed genes. Also, the variability preservation allowed the identification of new differentially expressed genes. Using the proposed approach, we have identified differentially expressed genes involved in pathways in cancer and confirmed some molecular markers already reported in the literature.
Highlights
The growing number of omics datasets and the recent advances in multi-omics integration approaches have contributed to the better understanding of biological mechanisms and the emergence of the personalized medicine
Microarrays have been predominantly used since the end of the nineties to measure gene expression levels, they remain widely used to detect other omics data types, including microRNA expression, DNA methylation, single-nucleotide polymorphisms (SNPs), and copy number variants (CNVs) (Goodwin et al, 2016)
In the section Intestinal Metaplasia Database, we describe a gene expression dataset used to illustrate the application of our proposed methods
Summary
The growing number of omics datasets (e.g., genomics, transcriptomics, proteomics, metabolomics) and the recent advances in multi-omics integration approaches have contributed to the better understanding of biological mechanisms and the emergence of the personalized medicine. Lee et al (2008) proposes to select the smoothing parameter by minimizing the bootstrap estimate of the mean integrated square error (MISE) and show that their results are comparable to OLIN All these methods have shown superiority over LOWESS normalization with a fixed arbitrarily chosen smoothing parameter, they lack in taking into account any heteroskedasticity in the data. Results show that a more robust and conservative analysis is performed when the LOWESS smoothing parameter is selected by our method, potentially reducing the number of false-positive differential expressions Both the pixel-level variabilities incorporated by the proposed estimators for the Mt and At values and the variability preserved by our more parsimonious normalization method contributed to the identification of new differentially expressed genes.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
