Run scRNA-seq data analysis pipeline + visualization

Single cell RNA sequencing (scRNA-seq) is a genomic approach used to quantify messenger RNA sequences (whole transcriptome) at the resolution of a single cell, which can be scaled to tens of thousands of cells at the same time. This results in generation of a cell by gene counts matrix (in R gene by cell) as the primary data, and additional metadata from any given scRNA-seq experiment. Data transformation, modelling, and analysis methods operate downstream on the above described data. scPipeVis is an R package that lets you perform all the steps with a single function so users can download the package and run this function to do a very preliminary bare-bones data analysis of scRNA-seq data. Additionally there is a visualisation function that creates basic plots which are typically used for the visualisation of scRNA-seq data. The complete motivation behind the creation of this package follows. There are major limitations of this package which are also talked about below.

A default single cell RNA sequencing data analysis pipeline comprises of several steps, from quality control, normalization of count matrix, feature selection, dimensionality reduction, clustering, annotating cell types, and then differential gene expression followed by gene set enrichment analysis. In my literature review, I found that all of these steps can be performed in so many different ways by using different methods at each step. This process is further complicated by tuning of hyperparameters at many different steps. Therefore, any user while analyzing their data, has to make so many choices/decisions at each step. These decisions, like tuning the hyperparameter or choosing the right method are absolutely not trivial. In other words, these choices at each step can affect your downstream analyses. Therefore, making well informed decisions are crucial. This is where scPipeVis comes in, following an extensive literature review (Amezquita et al. 2019), I have put together a pipeline that takes care of most of these above mentioned steps using the current best practices methods based on published reviews benchmarking different methods for all these steps. Further scPipeVis provides a function to visualize each step of the pipeline with many different and appropriate depictions. These depictions are chosen after going through the literature and figuring out what kind of visualizations are required at each step. Therefore, this part of the package will facilitate informed decision making for users as they go on to select the best suitable methods and hyper parameters which in turn best suits their own datasets and meets their needs. The aim is to reduce the initial decision burden and provide a starting point for further, more specialised analyses.

This package is created for preliminary data analysis of scRNA-seq data, at the end of the day the user will have to eventually figure out if certain methods work better to answer the question that they are trying to answer using their data. The user will also eventually have to tune hyperparameters of certain functions, if they dont get satisfactory results from this pipeline. Last but not the least, here I don’t provide an exhaustive set of visualizations. I will continue working on the package to add more and more visualizations and make the pipeline more end to end.

scPipeVis package was developed using R version 4.5.1 (2025-06-13), Platform: aarch64-apple-darwin20 and Running under: macOS Sonoma 14.3

To install the latest version of scPipeVis:

install.packages("devtools")
library("devtools")
devtools::install_github("ArhanUofT/scPipeVis", build_vignettes = TRUE)
library("scPipeVis")

To run the Shiny app:

run_scPipeVis_app()

ls("package:scPipeVis")
browseVignettes("scPipeVis")

scPipeVis currently provides four main user-facing functions:

• run_pipeline()

  Run a standard scRNA-seq analysis pipeline on a SingleCellExperiment object. The pipeline includes quality control metrics, log-normalisation, highly variable gene selection, PCA, UMAP, and graph-based clustering, with results stored back into the SingleCellExperiment.

• visualise_pipeline()

  Generate diagnostic plots summarising major pipeline steps, including the mean–variance relationship used for feature selection, low-dimensional embeddings (PCA and UMAP), QC distributions, and simple cluster-level summaries. Plots are returned as a named list of ggplot objects and are also combined and saved as a single multi-panel figure.

• plot_gene_on_umap()

  Visualise expression of a single gene on the UMAP embedding using a viridis colour scale. This is useful for quick marker gene exploration. This can be useful for manual annotations or validating cluster annotations using marker genes.

• run_scPipeVis_app()

  Launch an interactive Shiny app which runs the pipeline (if needed) and allows the user to select any gene and visualise its expression on the UMAP embedding via a simple graphical interface. Please read the functions man page to understand how to pass in your own data to the shiny app.

Refer to package vignettes for more details. An overview of the package is illustrated below.

• scPipeVis was developed by Arhan Rupani, 4th year Bioinformatics and Computational Biology student at the University of Toronto. The author did an extensive literature review to develop a basic understanding of the field of scRNA-seq data analysis and created this pipeline.
• The run_pipeline function was written by the author and runs a typical scRNA-seq pipeline given the counts matrix/data in a SingleCellExperiment object. The visualise_pipeline fucntion was written by the author and creates typical plots which are used during exploratory scRNA-seq data analysis. The plot_gene_on_umap function was written by the author and is inspired by the need to regularly visualise expression of marker genes after clustering for manual annotations or checking annotations. This function was originally needed for the authors other research project and found its way into this package. The shiny app is mainly build around this function as it makes for a perfect use case for the shiny app. All the functions make use of functions implemented in already existing R packages made for scRNA-seq data analysis. These packages are mentioned below and their full references are in the References section below. There are intext citations present in the man pages for all the functions wherever an already existing function from another package was used.
• Packages used: SingleCellExperiment (Lun et al. 2016), scuttle (McCarthy et al. 2017), scater (McCarthy et al. 2017), scran (Lun et al. 2016), S4Vectors, ggplot2
• Generative AI was used for assignment 5 for coding help only. However, generative AI was not used until assignment 4 and the descison to use it in assignment 5 was made to not be at a disadvantage compared to other students in the class.

• Amezquita RA, Lun ATL, Becht E, Carey VJ, Carpp LN, Geistlinger L, Marini F, Rue-Albrecht K, Risso D, Soneson C, et al. 2019 Dec 2. Orchestrating single-cell analysis with Bioconductor. Nature Methods. doi:https://doi.org/10.1038/s41592-019-0654-x.
• Hadley Wickham. 2016. ggplot2 Elegant Graphics for Data Analysis. Cham Springer International Publishing.
• Hadley Wickham. 2019. Advanced R. Boca Raton: Chapman & Hall/Crc.
• Lun ATL, McCarthy DJ, Marioni JC. 2016. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Research. 5:2122. doi:https://doi.org/10.12688/f1000research.9501.2.
• McCarthy DJ, Campbell KR, Lun ATL, Wills QF. 2017 Jan 14. Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics.:btw777. doi:https://doi.org/10.1093/bioinformatics/btw777.
• Risso D, Cole M (2025). scRNAseq: Collection of Public Single-Cell RNA-Seq Datasets. doi:10.18129/B9.bioc.scRNAseq, R package version 2.23.1, https://bioconductor.org/packages/scRNAseq.
• Wickham H. 2015. R Packages. “O’Reilly Media, Inc.”
• Slides for BCB410 by professor Anjali Silva
• Chang W, Cheng J, Allaire J, Sievert C, Schloerke B, Aden-Buie G, Xie Y, Allen J, McPherson J, Dipert A, Borges B (2025). shiny: Web Application Framework for R. R package version 1.12.1, https://shiny.posit.co/.

This package was developed as part of an assessment for 2025 BCB410H: Applied Bioinformatics course at the University of Toronto, Toronto, CANADA. scPipeVis welcomes issues, enhancement requests, and other contributions. To submit an issue, use the GitHub issues.