M3Drop: dropout-based feature selection for scRNASeq

By Tallulah S. Andrews, Martin Hemberg. 2018

236 citations recorded by the source corpus.

Abstract Motivation Most genomes contain thousands of genes, but for most functional responses, only a subset of those genes are relevant. To facilitate many single-cell RNASeq (scRNASeq) analyses the set of genes is often reduced through feature selection, i.e. by removing genes only subject to technical noise. Results We present M3Drop, an R package that implements popular existing feature selection methods and two novel methods which take advantage of the prevalence of zeros (dropouts) in scRNASeq data to identify features. We show these new methods outperform existing methods on simulated and real datasets. Availability and implementation M3Drop is freely available on github as an R package and is compatible with other popular scRNASeq tools: https://github.com/tallulandrews/M3Drop. Supplementary information Supplementary data are available at Bioinformatics online.

Subfields: Molecular Biology

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

Feature selection and dimension reduction for single-cell RNA-Seq based on a multinomial model (2019, background): “Alternatively, highly dropout genes may be retained [18].”
Integrated analysis of multimodal single-cell data (2020, unknown): “Inspired by the M3Drop (Andrews and Hemberg, 2019), we identified ‘variable’ features in a group of cells based on unexpectedly high ‘dropout’ rates after controlling for a gene’s average expression.”
Orchestrating single-cell analysis with Bioconductor (2019, unknown): “50,51 for reviews in feature selection methods.”
Probabilistic harmonization and annotation of single‐cell transcriptomics data with deep generative models (2019, unknown): “As the scmap manuscript insists heavily on why the M3Drop (Andrews & Hemberg, 2019) gene filtering procedure is crucial to overcome batch effects and yield accurate mapping, we let scmap choose its default number of genes ( g = 500) with this method.”
A comparison of automatic cell identification methods for single-cell RNA sequencing data (2019, result): “Using genes as features that have a higher number of dropouts compared to the expected number of dropouts has been shown to yield the best results [17, 18].”
Droplet scRNA-seq is not zero-inflated (2019, background): “One group proposed that if data is assumed to follow negative binomial noise, genes with excessive zeros might be good candidates for further analysis (Andrews and Hemberg, 2018).”
Comparison and evaluation of statistical error models for scRNA-seq (2021, methodology): “To evaluate the procedure’s effectiveness, we employed a metric inspired by [44] and calculated the overlap of highly variable genes with the list of marker genes identified using unsupervised clustering analysis (see the “Methods” section).”