Robust methods for differential abundance analysis in marker gene surveys

By Joseph N. Paulson, O. Stine, H. Bravo, Mihai Pop. Nature Methods, 2013

1,918 citations recorded by the source corpus.

We introduce a methodology to assess differential abundance in sparse high-throughput microbial marker-gene survey data. Our approach, implemented in the metagenomeSeq Bioconductor package, relies on a novel normalization technique and a statistical model that accounts for undersampling—a common feature of large-scale marker-gene studies. Using simulated data and several published microbiota data sets, we show that metagenomeSeq outperforms the tools currently used in this field.

Subfields: Molecular Biology

View Paper: https://www.semanticscholar.org/paper/a6e0739455352b0be05e22548bff81f5b1f80788

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

Soil Bacterial and Fungal Response to Wildfires in the Canadian Boreal Forest Across a Burn Severity Gradient (2019, positive): “unburned plots) using metagenomeSeq (Paulson et al., 2013), after controlling for (including as variables) vegetation community (categorical variable), pH (continuous variable), and %C (continuous variable), resulting in an estimate of the log2-fold change in the abundance of each OTU in burned vs.”
Effects of library size variance, sparsity, and compositionality on the analysis of microbiome data (2015, methodology): “Also, the addition of a constant (or 408 pseudocount; here, one) to the count matrix prior to CSS (Paulson et al. 2013), DESeq (Anders 409 & Huber 2010), and logUQ (Bullard et al. 2010) transformation as a way to avoid log(0) is not 410 ideal, and clustering results have been shown to be very…”
The Hidden World within Plants: Ecological and Evolutionary Considerations for Defining Functioning of Microbial Endophytes (2015, methodology): “The assigned KO tags were normalized by cumulative sum scaling (CSS) normalization, and then a mixture model that implements a zero-inflated Gaussian distribution was computed to detect differentially abundant properties by using the metagenomeSeq package (249).”
Combined amendment of immobilizers and the plant growth-promoting strain Burkholderia phytofirmans PsJN favours plant growth and reduces heavy metal uptake (2015, methodology): “A data-driven adaptive method for selecting normalization scale quantile was conducted and data normalized by scaling counts by the nth percentile of each sample's nonzero count distribution in the metagenomeSeq Bioconductor package (Paulson et al., 2013; McMurdie and Holmes, 2014).”
A Novel Sparse Compositional Technique Reveals Microbial Perturbations (2019, background): “This is particularly useful when one assumes that the data are lognormally distributed as proposed in reference 42, since log-transformed lognormally distributed data are normally distributed.”
LinDA: linear models for differential abundance analysis of microbiome compositional data (2021, methodology): “The normalizing factor could also be included as an offset in regression models such as edgeR [15], DESeq2 [16], MicrobiomeDDA [17], and metagenomeSeq [13], where the TMM, RLE, GMPR, and CSS normalization are the accompanying normalization methods.”
QIIME 2 Enables Comprehensive End‐to‐End Analysis of Diverse Microbiome Data and Comparative Studies with Publicly Available Data (2020, neutral): “To ensure a successful data analysis, begin metadata generation at the time of sample collection.”
Analysis of composition of microbiomes: a novel method for studying microbial composition (2015, methodology): “Our extensive simulation studies show that ANCOM outperforms Zero Inflated Gaussian (ZIG) methodology (1) by substantially reducing the FDR and increasing power.”
Correlation detection strategies in microbial data sets vary widely in sensitivity and precision (2016, background): “Rarefying, or drawing without replacement from each sample’s distribution until all samples have the same total number of sequences, metagenomeSeq’s cumulative sum scaling (Paulson et al., 2013) and DESeq’s log-ratio-based variance stabilizing transformation (Anders and Huber, 2010) were examined.”