VLP Base Editor Collection Benchmarking

Data analysis pipeline for benchamrking base editors delivered via engineered virus-like particles (VLPs) across three genomic targets (HEK3, B2M, PDCD1), six VLP doses (D0 = 0 µL – D5 = 10 µL), and three replicates. Raw amplicon sequencing reads are downloaded from SRA, merged, QC'd, demultiplexed by replicate barcode, quantified with CRISPResso2, and summarized into editing/substitution/deletion tables and figures.

Pipeline overview

The pipeline runs as four separate Snakefiles, in order, followed by an R plotting script:

Step	File	What it does
0	workflow/Snakefile_sra_download	Downloads raw FASTQs from SRA into data/seq_data/ using config/sra_accession_map.tsv
1	workflow/Snakefile_Fastqc	Raw-read FastQC, merges paired-end reads with pear, MultiQC report
2	workflow/Snakefile_demux	Splits each merged per-sample FASTQ into per-replicate FASTQs via 5′-anchored barcode matching (cutadapt)
3	workflow/Snakefile_crispresso	Runs CRISPRessoPooled per sample × replicate, then CRISPRessoAggregate per base editor, and builds a combined editing summary table
4	workflow/scripts/crispresso_analysis.R	Loads the CRISPResso output tables and produces all summary figures (PDF)

See each subfolder's own README for details:

• config/ — sample sheet, SRA accession map, barcode map, amplicon/target definitions, pipeline parameters
• workflow/ — Snakefiles and helper scripts
• data/ — raw sequencing data
• results/ — pipeline outputs

Setup

Two conda environments are used, plus sra-tools for the download step:

conda env create -f envs/be_analysis.yml

Running the pipeline

Run each Snakefile from the repository root, activating the matching environment:

conda activate be_analysis

# 0. Download raw reads from SRA
snakemake -s workflow/Snakefile_sra_download --cores 8

# 1. QC + read merging
snakemake -s workflow/Snakefile_Fastqc --cores 8

# 2. Replicate demultiplexing
snakemake -s workflow/Snakefile_demux --cores 8

# 3. CRISPResso quantification
snakemake -s workflow/Snakefile_crispresso --cores 8

Add -n to any command to do a dry run first. Paths and parameters (raw data location, barcode error rate, CRISPResso quantification window, etc.) are set in config/config.yaml.

Once results/crispresso_tables/ is populated, generate the figures with:

# from within results/crispresso_tables/ — see workflow/scripts/README.md
# for the exact folder layout this script expects
source("../../workflow/scripts/crispresso_analysis.R")

results/crispresso_tables/ contains the curated set of CRISPResso output tables, so the figures can be regenerated directly without rerunning steps 0–3.