Data-adaptive pipeline for filtering and normalizing metabolomics data

Abstract

IntroductionUntargeted metabolomics datasets contain large proportions of uninformative features and are affected by a variety of nuisance technical effects that can bias subsequent statistical analyses. Thus, there is a need for versatile and data-adaptive methods for filtering and normalizing data prior to investigating the underlying biological phenomena. ConclusionOur proposed data-adaptive pipeline is intuitive and effectively reduces technical noise from untargeted metabolomics datasets. It is particularly relevant for interrogation of biological phenomena in data derived from complex matrices associated with biospecimens.

Denoising Search doubles the number of metabolite and exposome annotations in human plasma using an Orbitrap Astral mass spectrometer

“…Second, electronic noise becomes relatively more prominent for MS/MS spectra that originated from very low-intensity precursor ions. Therefore, a simple cut-off threshold at 1% base peak intensity does not suffice for metabolomics or exposome nontargeted studies that aim at low abundant molecules 20,22 .…”

“…Consequently, existing denoising methods are unsuitable for large scale, standardized de-noising in metabolomics. Typical metabolomics data processing software denoises spectra by simply discarding ions below 0.5-1% of the basepeak height [20][21][22] . Surprisingly, the chemistry information revealed by the fragment peaks are not considered when determining if a given fragment is true ion or noise.…”

The Hitchhiker’s Guide to Statistical Analysis of Feature-based Molecular Networks from Non-Targeted Metabolomics Data

“…But this is a tedious process 7 ; there is also Lawson et al's msPurity R package with a function called "SubtractMZ" to perform blank removal 110 . Data-adaptive filtering methods have also been suggested to remove features from blanks and low abundant features from samples with undetected values 111 . Another popular feature filtering method is based on the Coefficient of Variance (CV).…”

Metabolomics of neonatal blood spots reveal distinct phenotypes of pediatric acute lymphoblastic leukemia and potential effects of early-life nutrition

“…Then, acetonitrile was added to precipitate proteins and extracts were analyzed by LC-HRMS [8]. Data processing was performed in the R statistical programming environment using methods described elsewhere [15]. Detection of sample outliers, beyond a proportional expansion value of 1.2 for Hotelling's ellipse (PC1 and PC2), resulted in removal of two cases and 10 controls.…”

“…Feature selection was performed separately for early-and late-diagnosis of ALL. Peak areas were log transformed and normalized with the Bioconductor R package 'scone' [15,19,20], which implemented and evaluated different scaling and regression-based-normalization methods for removing unwanted variation while preserving differences in case status. The normalization scheme selected by 'scone' used DESeq scaling and accounted for the following unwanted sources of variation: NBS age, blood volume, run order, and batch.…”