OpenSplice

Overview

OpenSplice quantifies the impact of >590,000 variants on the alternative splicing of 608 human exons by massively parallel site-saturation mutagenesis in minigene constructs. This repository contains all code to reproduce the analyses and figures in the paper.

Interactive browser ExonExplorer: https://results.hgi.sanger.ac.uk/OpenSplice/

---

Repository structure

OpenSplice/
├── OpenSplice.Rproj                    ← open this in RStudio
│
├── libraries_design/                   ← library design scripts (run once, before sequencing)
│   ├── 01_wt_screening_libraries/      ← exon selection for the 6k WT screen
│   └── 02_mutagenesis_libraries/       ← saturation mutagenesis oligo design
│
├── psi_calculation_pipeline/           ← HPC sequencing processing pipeline
│   ├── 01_bc_var_association/          ← barcode–variant association (bash + Python)
│   ├── 02_dimsum/                      ← read count aggregation via DiMSum
│   ├── 03_psi_per_barcode/             ← PSI per barcode (bash + Python + R)
│   └── 04_psi_per_variant/             ← aggregate + error model + normalisation (R)
│
├── analysis/                           ← all R analyses and figures (see analysis/README.md)
│   ├── config.R                        ← central path hub — sourced by every script
│   ├── 00_master_table_creation.R      ← build master table from pipeline outputs
│   ├── 01.x – 07_*.R                   ← analysis scripts (see analysis/README.md)
│   └── README.md
│
├── data/
│   ├── raw/                            ← raw sequencing data 
│   ├── processed/                      ← HPC pipeline outputs (add here the PSI per barcode)
│   └── databases/                      ← external reference files (MaxEntScan, ClinVar, …)
│
├── results/
│   ├── psi_per_variant/                ← output of 'psi_calculation_pipeline/04_psi_per_variant/'
│   ├── analysis/                       ← per-script result tables
│   └── supplementary_tables/           ← Supplementary Tables (TSV)
│
├── figures/                            ← stored figures
│
└── envs/
    ├── requirements.txt                ← Python dependencies (pip / conda)
    └── session_info.R                  ← capture R package versions for reproducibility

---

Quick start

1. Clone and open the project

git clone https://github.com/lehner-lab/OpenSplice.git
cd OpenSplice

Open OpenSplice.Rproj in RStudio. The here package resolves all paths automatically from the project root — no manual path editing is needed.

2. Install R dependencies

install.packages("renv")
renv::restore()

3. Download data

Download raw and processed data from ENA (PRJEB111846) and place them in data/raw/ Download PSI per barcode tables from Figshare (https://doi.org/10.6084/m9.figshare.32337414) and place them in data/processed/03_psi_per_barcode

The following external database files must be downloaded separately and placed in data/databases/ (paths are defined in analysis/config.R):

File	Source
SpliceVarDB/20250224_splicevardb.download.tsv	splicevardb.org
other_dms/GSE307247_Processed_PSIs_All_Cells.csv	GEO GSE307247 — file GSE307247_Processed_PSIs_All_Cells.csv.gz, decompress before use
branch_point/ — Mercer et al. 2015 Supplementary Table 1	Genome Research
branch_point/ — Taggart et al. 2017 Supplementary Table 5	Genome Research
branch_point/lstm.gencode_v19.hg19.top.bed.gz	LaBranchoR — http://bejerano.stanford.edu/labranchor/downloads/dat/lstm.gencode_v19.hg19.top.bed.gz
branch_point/lstm.gencode_v19.hg19.all.tsv.gz	LaBranchoR — http://bejerano.stanford.edu/labranchor/downloads/dat/lstm.gencode_v19.hg19.all.tsv.gz

4. HPC pipeline

The psi_calculation_pipeline/ directory contains the compute-intensive processing steps (barcode–variant association → DiMSum → PSI per barcode → PSI per variant). See psi_calculation_pipeline/README.md for cluster-specific instructions.

Shortcut 1: if you download the psi_per_barcode files from Figshare (https://doi.org/10.6084/m9.figshare.32337414) and place them under data/processed/03_psi_per_barcode/, you can skip steps 01–03 and start directly from psi_calculation_pipeline/04_psi_per_variant/.

Shortcut 2: if you download the Supplementary_Table4.tsv file from the preprint and place it under results/supplementary_tables/Supplementary_Table4.tsv, you can skip all the psi_calculation_pipeline steps + analysis/00_master_table_creation.R and start directly from analysis/01.1_replicates_correlation_plots.R.

5. Build the master table

Run once to produce results/supplementary_tables/Supplementary_Table4.tsv, which is read by every downstream analysis script:

source("analysis/00_master_table_creation.R")

6. Run analyses

Each numbered script in analysis/ is self-contained. Run in order:

source("analysis/01.1_replicates_correlation_plots.R")
source("analysis/01.2_other_dataset_comparison.R")
# ...
source("analysis/07_clinvar.R")

See analysis/README.md for a full description of every script.

---

Data availability

• DNA/cDNA sequencing data — European Nucleotide Archive (ENA) under accession PRJEB111846.
• PSI values per variant — Supplementary Table 4 of the paper.
• Processed prediction scores — Supplementary Table 12.
• Barcode-level read counts, exon/variant sequences, flanking genomic regions, and unprocessed predictor scores — Figshare: https://doi.org/10.6084/m9.figshare.32337414

---

Citation

Quarantani G, Clarke J, Thompson M, Sang F, Valcárcel J, Lehner B. OpenSplice: the impact of half a million mutations on the alternative splicing of 600 human exons. bioRxiv 2026. https://doi.org/10.64898/2026.05.22.727141