henhaobio
Sign in

bioRxivBioinformaticsPreprint

Generating publication ready visualizations for Single Cell transcriptomics using SCpubr

Enrique Blanco-CarmonaiD

PDFDOI
In-depth full-text digest available · Jump to digest ↓

Abstract

Single Cell transcriptomic analysis has become a widespread technology of choice when it comes to understanding the differences at a transcriptomic level in heterogeneous samples. As a consequence, a plethora of analysis tools have been published to tackle the different analysis steps from count matrix generation to downstream analysis. Many of them provide ways to generate visualizations of the data. While some design choices are made, it is a common practice to provide the user with visualizations as raw as possible so that they can be customized to the user needs. However, in many cases these final customization steps are either time consuming or demand a very specific set of skills. This problem is addressed by SCpubr, which sacrifices some of this initial freedom of choice in aesthetics to provide the user a more streamlined way of generating high quality Single Cell transcriptomic visualizations.

In-depth summary

Claude Opus 5.5 · xhigh effort · 17 Oct 2026 · from the full text of version 1 (1 Mar 2022) · CC BY-NC 4.0

An R package that restyles the standard Seurat single-cell plots with fixed defaults, keeps the whole embedding visible when panels are split, and adds bar and bee swarm options.

Analysis packages such as Seurat deliberately return plots in a raw, customisable state, so getting figures to publication quality costs time or a specialised skill set. SCpubr gives up some of that freedom and applies opinionated defaults instead, wrapping the usual functions behind a Seurat-object interface.

Key results

  1. ClaimedWhen a plot is split by a grouping variable, the cells outside each panel's group are kept as grey points, so the shape of the full embedding is not lost, and a feature plot keeps one colour gradient across panels together with its legend.

    The author contrasts this with Seurat::DimPlot and Seurat::FeaturePlot, where splitting separates cells into panels and the overall UMAP outline disappears, and where the shared scale across split feature panels comes without a legend showing the value range. No measurement or user comparison accompanies the claim, and no figures appear in the extracted text. · Dimensional reduction plot; Feature plots

  2. ClaimedMany functions take the output of an existing plotting function and restyle it rather than redrawing it, applying a fixed set of defaults across plot types.

    Many functions take another package's plot object and modify it with ggplot2, ggpubr, patchwork and colortools; do_VlnPlot, for example, starts from Seurat::VlnPlot output, drops the points and adds an inner boxplot plus an optional quality-control cutoff line. Shared defaults include no axes on UMAP, bold legends, a viridis gradient for continuous values, and shuffled cells with larger dots. · Dimensional reduction plot; Feature plots; Violin plots; Methods

  3. ClaimedThe bar and bee swarm functions go beyond restyling: bar plots for cell-type composition and rank-based bee swarm plots for continuous variables such as pseudotime.

    do_BarPlot draws one variable as plain bars or subdivides them by a second variable, as absolute counts or within-bar proportions, orderable by descending count or by a chosen subgroup's proportion, vertical or horizontal. do_BeeSwarmPlot, built on ggbeeswarm, ranks cells by a continuous variable, plots them separated by user-chosen groups and colours them by group or by the ranked value. · Bar plots; Bee swarm plots

Methods, data and reproducibility

An R package whose functions take a Seurat object and return a styled plot; some figures are drawn from scratch, many wrap another package's plot object and modify it with ggplot2, ggpubr, patchwork and colortools. The examples come from one public blood dataset processed in Seurat.

  • Public 10x Genomics blood mononuclear cell dataset: cellranger v6.1 counts, filtered at 1000 UMIs, 500 genes and 20% mitochondrial RNA, then RNBR normalisation, PCA, UMAP on 30 PCs, Louvain clusters · human · n = 10000 cells
Methods & tools
SCpubr, Seurat v4, cellranger v6.1, Nebulosa, ggbeeswarm, ggplot2, ggpubr, patchwork, colortools
Reagents & environment
R package designed for Seurat objects; the example analysis used cellranger v6.1 and Seurat v4. No SCpubr version or dependency versions are given in the text.
Reproducibility gaps
The text points to an installation source and a reference manual, but both addresses were lost in extraction, so no link can be listed; a CRAN release is described as planned rather than done. The demonstration dataset is named without an accession or link.
Limitations (4)
  • The package deliberately fixes aesthetic choices, so users trade some control for speed, though the author says further customisation remains possible. · stated by the authors
  • Distribution is not yet through CRAN; release there is described only as planned. · stated by the authors
  • No evaluation is reported: no timing, no user study, and no comparison beyond the author's description of what the wrapped functions do; no figures appear in the extracted text. · noted in this summary
  • Scope is limited to Seurat objects in R, so Python workflows such as scanpy are not served, and all examples come from a single 10000-cell blood dataset. · noted in this summary

Read this if you prepare single-cell figures from Seurat objects in R and want opinionated defaults instead of hand-tuning ggplot2 for every panel.

Citations
107

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISWang '19Speir '21Çakır '20Tabaka '19Kotliar '21Qing '25
69% semantic matchBioinformatics
Shuoguo Wang, Constance Brett, Mohan T. Bolisetty+3 more·bioRxiv·Jun 12, 2019· 11 citations