macpie

Introduction

macpie is a R toolkit designed for researchers, originally with MAC-seq data in mind, but validated for general High-Throughput Transcriptomics (HTTr) data applications. Its primary aim is to deliver the latest tools for quality control (QC), visualization, and analysis.

For processing raw sequencing data into count matrices, please refer to our companion Nextflow workflow:
dinoflow: Nextflow workflow for MAC-seq

Documentation

Full documentation and step-by-step tutorials are available at:
macpie documentation site

Example Data

We provide both full and subset datasets for testing and exploration:

• Full example dataset is hosted on Zenodo:
  https://doi.org/10.5281/zenodo.15778812

• Quick-start subset demo is bundled with the package and can be loaded directly in R:

data("demo", package = "macpie")

Requirements

• R ≥ 4.3.3

  The package is developed on R 4.3.3, and installation (including all dependencies) has been tested on the latest R 4.5.0.

Installation and dependencies

All required R packages are automatically installed:

• via pak::pkg_install or devtools::install_github

• or use our pre-built Docker image for a ready-to-use environment.

Installing locally

To install the development version of macpie, we recommend using pak - a fast package installer to install directly from GitHub. Make sure you have the pak package installed first.

Configure your own GitHub PAT (optional)

pak uses the GitHub API to resolve versions and fetch metadata—and unauthenticated requests are limited to 60 per hour. To avoid rate-limit errors, you can add your own Personal Access Token (PAT):

# install gitcreds if you haven’t already
install.packages("gitcreds")

# this will prompt you to paste in your PAT
gitcreds::gitcreds_set()

# 1. Install pak (fast installer)
if (!requireNamespace("pak", quietly=TRUE)) {
  install.packages("pak")
}

# 2. Make sure Bioconductor repos are set
if (!requireNamespace("BiocManager", quietly = TRUE)) {
  install.packages("BiocManager")
}
options(repos = BiocManager::repositories())

# 3. Re-install macpie + exactly its Depends & Imports via pak
pak::pkg_install(
  "PMCC-BioinformaticsCore/macpie",
  dependencies = c("Depends", "Imports")
)

# 4. Verify
# Should load without error:
library(macpie)

Optially, you can also install the package by

# First, install devtools if not already installed
install.packages("devtools")

# Make sure BiocManager is installed, and point your repos at both CRAN + Bioconductor
if (!requireNamespace("BiocManager", quietly=TRUE))
  install.packages("BiocManager")

options(repos = BiocManager::repositories())

BiocManager::install(c(
  "edgeR", "limma", "Biobase", "DESeq2", "RUVSeq",
  "EDASeq", "fgsea", "scran", "glmGamPoi",
  "BiocParallel", "SingleCellExperiment", "zinbwave",
  "SummarizedExperiment"
))


# Install MOFA2 from its GitHub (it’s not on Bioconductor)
devtools::install_github("bioFAM/MOFA2")

# Finally, install macpie itself
devtools::install_github("PMCC-BioinformaticsCore/macpie", dependencies = TRUE)

Using Docker image

Have your docker desktop running, open a terminal, paste the docker pull command and install, depending on your platform.

1. Pull the Docker image

docker pull --platform linux/amd64 xliu81/macpie:v1.0.0

2. Run the Docker container

docker run --rm -ti \
  -e PASSWORD=password \
  -p 8787:8787 \
  --platform linux/amd64 \
  -v /path/to/your/macpie/:/home/rstudio/macpie:z\
  xliu81/macpie:v1.0.0

• Replace /path/to/your/macpie/ with the absolute path to your local repo.

• Copy and paste http://localhost:8787 in your browser

  • Username: rstudio

  • Password: password (or the one you set in the docker run command)

After logging in, you’ll find your local directory mounted under:

/home/rstudio/macpie/

Quick start

In here we show using the demo dataset for a quick start. From each vignette on our website, we only include a couple of functions in this quick start.

library(macpie)

# load demo, 
# demo is a tidySeurat object with matched metadata
data("demo")

# Quality control
# Filter by counts per sample group
demo <- filter_genes_by_expression(demo,
                                  group_by = "combined_id",
                                  min_counts = 5,
                                  min_samples = 3)



# MDS plot
p <- plot_mds(demo, group_by = "Sample_type", label = "combined_id", n_labels = 30)
girafe(ggobj = p, fonts = list(sans = "sans"))


# Correction of the batch effect
# First we will subset the data to look at control, DMSO samples only
demo_dmso <- demo %>%
  filter(Treatment_1 == "DMSO")

# Run the RLE function
plot_rle(demo_dmso, label_column = "Row", normalisation = "limma_voom")


# Transcriptional analysis
# Single comparison
# First perform the differential expression analysis
treatment_samples <- "Staurosporine_10"
control_samples <- "DMSO_0"
top_table <- compute_single_de(demo, treatment_samples, control_samples, method = "limma_voom")

top_genes <- top_table %>%
  filter(p_value_adj < 0.1) %>%
  select(gene) %>%
  pull()

# A volcano plot with very small number of genes, as it's a subset of the full dataset 
plot_volcano(top_table, max.overlaps = 16)

# Multiple comparisons
# Filter out lower concentrations of compounds and untreated samples
treatments <- demo %>%
  filter(Concentration_1 == 10) %>%
  select(combined_id) %>%
  filter(!grepl("DMSO", combined_id)) %>%
  pull() %>%
  unique()
demo <- compute_multi_de(demo, treatments, control_samples = "DMSO_0", method = "limma_voom", num_cores = 1)

# plot shared differentially expressed genes
plot_multi_de(demo, group_by = "combined_id", value = "log2FC", p_value_cutoff = 0.01, direction="up", n_genes = 5, control = "DMSO_0", by="fc")

Citation

citation("macpie")

#> To cite the macpie package in publications, please use:
#>
#>  Bartonicek N, Liu X, Twomey L, Meier M, Lupat R, Craig S, Yoannidis D, Li J, Semple T,
#>  Simpson K, Li M, Ramm S (2025). “macpie: a scalable workflow for high-throughput
#>  transcriptomic profiling.” bioRxiv. doi:10.1101/2025.08.06.669002
#>
#> A BibTeX entry for LaTeX users is
#>
#>  @Article{,
#>    title   = {macpie: a scalable workflow for high-throughput transcriptomic profiling},
#>    author  = {Nenad Bartonicek and Xin Liu and Laura Twomey and Michelle Meier and Richard Lupat and Stuart Craig and David Yoannidis and Jason Li and Tim Semple and Kaylene J Simpson and Mark X Li and Susanne Ramm},
#>    journal = {bioRxiv},
#>    year    = {2025},
#>    doi     = {10.1101/2025.08.06.669002},
#>    url     = {https://www.biorxiv.org/content/10.1101/2025.08.06.669002v1}
#>  }