SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data.

Young MD, Behjati S. · 2020

Read SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. by Young MD, Behjati S. (2020) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc7763177.txt.

Introduction

Young and Behjati present a tool that cleans stray RNA out of single-cell data. Droplet-based methods read the RNA of single cells. But free RNA floats in the sample fluid and gets captured too. This background soup can distort what each cell seems to express. Its mix and size vary by experiment.

The tool, SoupX, measures how much contamination each sample holds. It then estimates each cell's true profile with the soup removed. Tests across several datasets and platforms showed cleaner biological signals after its use. Quality control scores also improved. The authors say the tool should raise the value of old and new datasets alike.

Catalog record

CREATORYoung MD, Behjati S.
YEAR2020
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC7763177

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data By Matthew D Young, Sam Behjati Published in: GigaScience DOI: 10.1093/gigascience/giaa151 ABSTRACT Abstract Background Droplet-based single-cell RNA sequence analyses assume that all acquired RNAs are endogenous to cells. However, any cell-free RNAs contained within the input solution are also captured by these assays. This sequencing of cell-free RNA constitutes a background contamination that confounds the biological interpretation of single-cell transcriptomic data. Results We demonstrate that contamination from this "soup" of cell-free RNAs is ubiquitous, with experiment-specific variations in composition and magnitude. We present a method, SoupX, for quantifying the extent of the contamination and estimating "background-corrected" cell expression profiles that seamlessly integrate with existing downstream analysis tools. Applying this method to several datasets using multiple droplet sequencing technologies, we demonstrate that its application improves biological interpretation of otherwise misleading data, as well as improving quality control metrics. Conclusions We present SoupX, a tool for removing ambient RNA contamination from droplet-based single-cell RNA sequencing experiments. This tool has broad applicability, and its application can improve the biological utility of existing and future datasets. INTRODUCTION Droplet-based single-cell RNA sequencing (scRNA-seq) has enabled quantification of the transcriptomes of hundreds of thousands of cells in single experiments [1,2]. This technology underpins recent advances in understanding normal and pathological cell behaviour [3–8]. Moreover, large-scale efforts to create a ”Human Cell Atlas" critically depend on the accuracy and cellular specificity of the transcriptional readout produced by droplet-based scRNA-seq [9,10]. A core assumption of droplet-based scRNA-seq is that each droplet, within which molecular tagging and reverse transcription take place, contains messenger RNA (mRNA) from a single cell. Violations of this assumption, which may distort the interpretation of scRNA-seq data, are common in practice. Clear examples include droplets that contain multiple cells (doublets), and empty droplets. Attempts to detect and remove doublets are an active area of research [11–13]. Another phenomenon that violates this assumption is the sequencing of cell-free RNA from the input solution, admixed with a cell in its enclosing droplet. It is recognized that these contaminating non-endogenous RNAs are present even within datasets of the highest quality [2]. Here, we show that this "soup" of cell-free mRNAs is ubiquitous and non-negligible in magnitude. Because the character and extent of ambient mRNA contamination varies by experiment, with increased contamination in necrotic or complex samples, ambient mRNAs may significantly confound the biological interpretation of scRNA-seq data. We present SoupX, a method for quantifying the extent of ambient mRNA contamination whilst purifying the true, cell-specific signal from the observed mixture of cellular and exogenous mRNAs. In this article we begin by briefly describing the SoupX method. Following this we consider a range of datasets, summarized in Supplementary Table S1. We first investigate 2 “species mixing” datasets run on the Chromium 10X [2] and DropSeq [14] platforms, which allow us to directly identify contaminating mRNAs and test our method’s accuracy. We then demonstrate how SoupX can be applied in practice using a dataset of peripheral blood mononuclear cells (PBMCs) [2]. We further explore the biological benefits of SoupX using a complex "kidney tumour" dataset, which consists of 12 kidney tumour biopsies [15]. As a final test, we apply our method to human fetal liver data [16]. We conclude with some general remarks about ambient RNA contamination, other tools to correct for its effect, and the consequences of failing to account for the presence of ambient RNAs. THE SOUPX METHOD Droplet-based scRNA-seq methods produce counts of unique molecular identifiers (UMIs) for genes in thousands of cells. The aim of an scRNA-seq experiment is to infer the number of molecules present for each type of gene within each cell from these data. However, the observed counts arise from a mixture of mRNAs produced by the captured cell and those present due to background contamination. SoupX aims to remove the contribution of the cell-free mRNA molecules from each cell and recover the true molecular abundance of each gene in each cell. The algorithm consists of the following 3 steps (summarized in Fig. 1): - Estimate the ambient mRNA expression profile from empty droplets. - Estimate (or manually set) the contamination fraction, the fraction of UMIs originating from the background, in each cell. - Correct the expression of each cell using the ambient mRNA expression profile and estimated contamination. SoupX produces a modified table of counts, which can be used in place of the original count matrix in any downstream analysis tool. [Figure: A visual summary of the SoupX method, using data from the PBMC dataset.] To estimate the background expression profile we consider all droplets with <Nemp UMIs, which we assume unambiguously do not contain cells. The fraction of background expression from gene g, bg, is then given by, (1)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{equation*} b_{g} = \frac{\sum _d n_{g,d}}{\sum _d \sum _g n_{g,d}}, \end{equation*}\end{document}where ng, d is the number of counts for gene g in droplet d and the sum over d is taken over all droplets with <Nemp UMIs (Fig. 1). The species-mixing experiment allows us to compare how accurately bg recapitulates the true background expression found within each cell, revealing that any value of Nemp < 100 produces a good correlation, with the best correlation given when Nemp < 10 (Supplementary Fig. S2). The most challenging part of using SoupX is estimating or specifying the number of UMIs in each cell that are contributed by background contamination. In general, the observed number of UMIs for gene g in cell c is given by (2)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{equation*} n_{g,c} = m_{g,c} + o_{g,c}, \end{equation*}\end{document}where mg, c are the cell endogenous counts and og, c are the counts from the background. We assume that the relative abundance of genes that make up the background does not differ between cells, which allows us to write, (3)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{equation*} o_{g,c} = N_c \rho _c b_g, \end{equation*}\end{document}where Nc = ∑gng, c, and ρc is the background contamination fraction. In general mg, c is unknown and what we are aiming to measure. To proceed, we assume that there is a combination of genes and cells for which mg, c = 0 exists. The genes for which mg, c = 0 for a given cell are those genes that are strong negative markers of the cell type c. For example, the gene HBB is a strong positive marker for erythroid cells (red blood cells) but should not be expressed in any other cell type. So for any cell c that is not an erythroid cell, HBB will not be expressed (i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $m_{HBB,\mathrm{not\ Erythroid}} = 0$\end{document}). Given a set of genes/cells for which we can assume that there is no cell endogenous expression (i.e., mg, c = 0) we calculate the cell-specific contamination fraction, (4)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{equation*} \rho _c = \frac{\sum _g n_{g,c}}{N_c \sum _g b_g}, \end{equation*}\end{document}where the sum is taken across all genes in cell c for which it is assumed mg, c = 0. SoupX optionally uses clustering information to refine the set of cells for

Read the complete text