IsoDiffR is a computational tool specifically designed to identify RNA isoforms whose expression patterns differ from those of their corresponding genes or major isoforms across distinct cell types, using single-cell RNA sequencing (scRNA-seq) data. It supports both pairwise and multi-cell-type comparisons, offering flexibility for diverse experimental designs.
The analysis of RNA isoforms in long-read single-cell RNA-seq data represents a cutting-edge direction in transcriptomics, providing insights that transcend conventional gene-level analyses. Despite the promise of this approach, dedicated computational methods remain scarce, underscoring an urgent need for methodological innovation in this rapidly evolving field.
To address this gap, we developed IsoDiffR as a versatile and accessible tool that facilitates isoform-level differential expression analysis. By enabling more nuanced exploration of transcriptional diversity and cell-type-specific splicing regulation, IsoDiffR contributes to a deeper understanding of gene expression complexity at single-cell resolution.
IsoDiffR can be installed via this command
if (!requireNamespace("devtools", quietly = TRUE)){
install.packages("devtools")
}
devtools::install_github("Eveqian98/IsoDiffR", build_vignettes = TRUE)
or
if (!requireNamespace("BiocManager", quietly = TRUE)){
install.packages("BiocManager")
}
BiocManager::install(version='devel')
BiocManager::install("DiffIsoR")
IsoDiffR is designed to work with the following required inputs:
- A Seurat object that contains:
- Isoform-level count matrix, stored in a dedicated assay (e.g.,
"isoform"). - Cell type annotations, stored as a column in the
meta.dataslot.
- Isoform-level count matrix, stored in a dedicated assay (e.g.,
- A GTF file that includes:
- Transcript IDs (
transcript_id) - Corresponding gene IDs (
gene_id)
- Transcript IDs (
Both the annotated Seurat object and the properly formatted GTF file are essential for running IsoDiffR and cannot be omitted.
The meta.data matrix of the Seurat object must include a column with cell type annotations.
> unique(isoform.0h@meta.data$cluster)
[1] "Corneal epithelial cells" "Basal cells" "Conjunctival cells" "Limbal stem cells"
IsoDiffR expects the GTF file to include the mapping between isoforms and their corresponding genes, with the attributes named "gene_id" and "transcript_id" respectively.
> gtf[sample(nrow(gtf), 5), ]
seqnames PacBio type start end strand transcript_id width gene_id
1343338 7 PacBio exon 137591288 137591387 + PB.58172.31(DCAF4~ISM) 99 DCAF4
1441262 9 PacBio exon 56276808 56276896 - PB.62348.202(PPIF~NIC) 88 PPIF
640775 19 PacBio exon 49142503 49142605 + PB.25747.221(EMP3~NNC) 102 EMP3
538191 17 PacBio exon 28938710 28938879 - PB.20816.480(KPNA3~ISM) 169 KPNA3
1318949 7 PacBio exon 94837377 94837458 - PB.55305.179(STRN3~NNC) 81 STRN3
The function getDEIso() returns the identified differentially expressed isoforms (DEIs) based on the input Seurat object and GTF file.
DEIs = getDEIso(seurat_obj = isoform.0h,gtf = gtf,subset_ident = unique(isoform.0h@meta.data$cluster),cluster_column = "cluster")
The returned result includes four data frames.
multi.geneDEIs compared with their corresponding genes across multiple cell typesmulti.majorDEIs compared with their corresponding major isoforms across multiple cell typespair.geneDEIs compared with their corresponding genes between two cell typespair.majorDEIs compared with their corresponding major isoforms between two cell types
The returned data frames contain various pieces of information In the multi.gene and multi.major data frames
isoformTranscript identifiers to each DEIgene_idThe gene identifier corresponding to each DEImajor_isoformThe major isoform corresponding to each DEI.ratio_rangeThe range of the isoform expression relative to the corresponding gene expression across clusters.adj.r.squaredAdjusted R-squared value of the DEI.
> DEIs$multi.gene[sample(nrow(DEIs$multi.gene), 5), ]
isoform gene_id major_isoform ratio_range adj.r.squared
963 PB.17036.10(KLF4~NNC) KLF4 PB.17036.3(KLF4~NNC) 0.093 -0.053
680 PB.23661.186(C18orf32~NNC) C18orf32 PB.23661.186(C18orf32~NNC) 0.188 -0.349
996 PB.11125.216(RTN4~FSM) RTN4 PB.11125.158(RTN4~NNC) 0.213 -0.259
545 PB.23991.6(MYL12A~FSM) MYL12A PB.23991.140(MYL12A~FSM) 0.075 -0.414
730 PB.10371.19(RHOB~FSM) RHOB PB.10371.19(RHOB~FSM) 0.136 -0.292
The pair.gene and pair.major data frames return several distinct items.
conditionThe two cell types between which the DEI is identifiedratio.1/ratio.2The proportion of the DEI's expression relative to the gene's total expression in each of the two cell typesratio_differenceThe difference in the proportion of the DEI's expression relative to the gene's total expression between the two cell typesr×adj.R²The product of the DEI's Pearson correlation coefficient and cosine similarity
> DEIs$pair.gene[sample(nrow(DEIs$pair.gene), 5), 4:8]
# A tibble: 5 × 5
condition ratio.1 ratio.2 ratio_difference `r×adj.R²`
<chr> <dbl> <dbl> <dbl> <dbl>
1 Basal cells_Conjunctival cells 0.401 0.232 0.169 -0.964
2 Conjunctival cells_Limbal stem cells 0.604 0.71 0.106 -0.997
3 Conjunctival cells_Limbal stem cells 0.519 0.639 0.12 -0.995
4 Basal cells_Conjunctival cells 0.373 0.505 0.132 -0.989
5 Basal cells_Limbal stem cells 0.42 0.213 0.207 -0.945
The function plotDEIso() generates a line plot showing the expression pattern of a specific isoform, along with its corresponding gene and major isoform, across specified cell types.
p <- plotDEIso(seurat_obj = isoform.0h,gtf = gtf,subset_ident = unique(isoform.0h@meta.data$cluster),cluster_column = "cluster",transcript_id = "PB.17036.10 (KLF4~NNC)")
multi.gene
pair.major
