Differential Expression Analysis with Long Read RNA-Seq Data

Software accompanying the paper “Differential Expression Analysis with Long Read RNA-Seq Data” by Liu Z., Song C., Wang Z., Fan Q., Yu J., Hao N., Chen S., Sun X., Ding H. (2026)

---

Overview

This R package implements a hurdle negative binomial generalized linear model (hurdle-NB GLM) for differential expression analysis of long-read RNA-Seq data. It provides:

• Data preparation and validation (prepareDGE()). Support for matrix, data.frame, DGEList, DESeqDataSet, or SummarizedExperiment inputs
• Estimation of size factors for normalization (sizeFactorsEst())
• Estimation of tag-wise dispersions using emperical priors (tagwiseEst())
• Gene-level differential expression testing using:
  • Likelihood Ratio Test (hurdle.LRT())
  • Wald Test (hurdle.Wald.Test())

The package is optimized for long-read sequencing data with high dropout rates, where zero inflation and overdispersion are common.

---

File Structure

File / Folder	Description
R/estimateSizeFactors.R	Functions for estimating normalization size factors for RNA-Seq counts.
R/estimateDispersion.R	Functions for prior estimation and tag-wise dispersion estimation.
R/hurdleNBTest.R	Functions for hurdle-NB differential expression testing (Wald & LRT).
DESCRIPTION	Package metadata, dependencies, and version information.
NAMESPACE	Exports, imports, and internal function definitions.
README.md	Package overview, usage, and installation instructions.

---

Installation

To install the stable release version of LRDE from Bioconductor:

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("LRDE")

To install the development version (currently required while under review):

if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("LRDE", version = "devel")

Example usage

# Load the package
library(LRDE)

# 1. Simulate a count matrix (negative binomial)
set.seed(123)
mat <- matrix(rnbinom(300, size = 5, mu = 5), nrow = 50)
grp <- factor(c("A", "A", "A", "B", "B", "B"))

# 2. Prepare the data object
y <- prepareDGE(mat, grp)

# 3. Estimate size factors for normalization
y <- sizeFactorsEst(y)

# 4. Estimate tag-wise gene dispersions
y <- tagwiseEst(y)

# 5a. Differential expression testing: Wald test
y <- hurdle.Wald.Test(y)

# 5b. Differential expression testing: Likelihood Ratio Test
y <- hurdle.LRT(y)

# 6. Access results
y$lrt_stats        # LRT statistics per gene
y$wald_stats       # Wald statistics per gene
y$p.values         # Corresponding p-values
y$tagwise.disp     # Estimated dispersions
y$zero_prob_matrix # Zero probabilities per group

---

Contact

For questions, comments, or collaborations, please contact:

• Hongxu Ding — hongxuding@arizona.edu
• Ziyang Liu — ziyang773@arizona.edu

---