Isosceles

Isosceles (Isoforms from Single-Cell; Long-read Expression Suite) is an R package dedicated to transcript detection and quantification from long reads, supporting both bulk RNA-Seq and scRNA-Seq technologies.

Kabza M., Ritter A., Byrne A., Sereti K., Le D., Stephenson W., Sterne-Weiler T. Accurate long-read transcript discovery and quantification at single-cell, pseudo-bulk and bulk resolution with Isosceles. Nat Commun 15, 7316 (2024). https://doi.org/10.1038/s41467-024-51584-3

Installation

Isosceles can be installed using the following commands:

install.packages(c("BiocManager", "devtools"))
BiocManager::install(c("scran", "scater", "uwot", "dittoSeq", "DEXSeq", 
                       "Nebulosa", "ggbio", "BiocStyle"))
devtools::install_github("Genentech/Isosceles", dependencies = TRUE, upgrade = TRUE,
                         INSTALL_opts = "--install-tests")

We found that some versions of Isosceles' dependencies don't work together well, which might cause problems with testing the package or building the vignettes. If you encounter such issues, re-installing certain packages might be helpful:

install.packages("irlba") 
devtools::install_github("powellgenomicslab/Nebulosa", upgrade = FALSE) 

Load the Isosceles package:

library(Isosceles)

Usage

You can follow along with our vignettes (Introduction to the Isosceles package, Mouse E18 brain data analysis) or the reference manual!

Best practices

• We recommend minimap2 for all long-read alignments.
• As Isosceles doesn't perform post-alignment splice junction correction, it is necessary to run minimap2 with the '--junc-bed' flag. The intron position BED file required by it can be easily created using the gtf_to_intron_bed function.
• The default settings of de novo transcript detection used by the bam_to_tcc function should work well for a eukaryotic transcriptome, but for the analysis of spike-in data, such as SIRVs, we recommend increasing the read count threshold (the min_read_count argument) to a higher value (e.g. 50).

Troubleshooting

You can check if the package works correctly by running its unit tests:

testthat::test_package("Isosceles")

In case of any problems, we recommend using the Isosceles Singularity image you can download from Zenodo or installing the package in a Docker container using a Bioconductor image.