NormalyzerDE: Online Tool for Improved Normalization of Omics Expression Data and High-Sensitivity Differential Expression Analysis

Abstract

Technical biases are introduced in omics data sets during data generation and interfere with the ability to study biological mechanisms. Several normalization approaches have been proposed to minimize the effects of such biases, but fluctuations in the electrospray current during liquid chromatography–mass spectrometry gradients cause local and sample-specific bias not considered by most approaches. Here we introduce a software named NormalyzerDE that includes a generic retention time (RT)-segmented approach compatible with a wide range of global normalization approaches to reduce the effects of time-resolved bias. The software offers straightforward access to multiple normalization methods, allows for data set evaluation and normalization quality assessment as well as subsequent or independent differential expression analysis using the empirical Bayes Limma approach. When evaluated on two spike-in data sets the RT-segmented approaches outperformed conventional approaches by detecting more peptides (8–36%) without loss of precision. Furthermore, differential expression analysis using the Limma approach consistently increased recall (2–35%) compared to analysis of variance. The combination of RT-normalization and Limma was in one case able to distinguish 108% (2597 vs 1249) more spike-in peptides compared to traditional approaches. NormalyzerDE provides widely usable tools for performing normalization and evaluating the outcome and makes calculation of subsequent differential expression statistics straightforward. The program is available as a web server at .

Dysregulation in mTOR/HIF-1 signaling identified by proteo-transcriptomics of SARS-CoV-2 infected cells

“…For quantification both unique and razor peptides were requested. Protein raw data abundance was first filtered for empty rows with in house script and quantile-normalize using R package NormalyzerDE 29 . Principal component analysis (PCA) was applied to explore sample-to-sample relationships.…”

Toward a Standardized Strategy of Clinical Metabolomics for the Advancement of Precision Medicine

“…Regarding computational solutions, the new version of the statTarget package with a user-friendly graphical interface integrating a QC-based random forests signal correction algorithm has been released to remove inter-and intra-batch variations in omics study [82]. NormalyzerDE, an R package, provides 12 different normalization methods for omics data to enable researchers the correct technical variation, and decide the most suitable strategy for particular datasets [83]. NOREVA is another powerful online tool that implements various normalization methods of which five well-established criteria were supported for the comparison of different normalization approaches [84].…”

OmicLoupe: Facilitating biological discovery by interactive exploration of multiple omic datasets and statistical comparisons

“…Statistical visualizations require columns with p-value, false discovery rate (FDR) corrected p-values, fold change (difference of means between the two compared groups) and average feature level. These values are provided by up-stream software such as NormalyzerDE [7] or R packages such as Limma [8] for most types of omics or DESeq2 [9], for RNA-seq expression data. After loading the data in the web interface, the visualizations can be accessed immediately.…”

“…They were subsequently log2-transformed and rolled up to protein level using an R-reimplementation (github.com/ComputationalProteomics/ProteinRollup) of the DanteR RRollup [26], using default settings and excluding proteins supported by a single peptide. Statistical contrasts between the two concentration levels were subsequently calculated using Limma [8] as provided by NormalyzerDE [7], and resulting p-values adjusted for multiple hypothesis testing using the Benjamini-Hochberg procedure [27].…”