tiling_screens_with_permutation

Statistical pipeline for detecting non-canonical ORFs (ncORFs) with differential essentiality signals in CRISPR tiling screens.

Canonical workflow

The publication-ready analysis entrypoint is:

• scripts/ncorf_tiling_analysis.R

This script runs a reproducible ORF-vs-parent-gene test using:

• empirical Bayes shrinkage (ashr)
• gene-specific permutation null distributions
• Benjamini-Hochberg FDR correction

Legacy notebooks and earlier scripts are preserved under notebooks/archive/ and scripts/archive/ for provenance, but are not the official execution path.

Repository structure

.
├── examples/                     # Small public fixtures for smoke tests
├── scripts/
│   ├── ncorf_tiling_analysis.R   # Canonical analysis pipeline
│   ├── preprocessing/            # Optional upstream mapping utilities
│   └── archive/                  # Legacy non-canonical scripts
├── notebooks/archive/            # Legacy exploratory notebooks
├── tests/smoke/                  # Reproducibility and static checks
├── docs/
│   ├── reproducibility.md
│   └── data_access.md
├── LICENSE
├── CITATION.cff
├── .zenodo.json
└── CHANGELOG.md

Requirements

R

• R >= 4.2
• packages: tidyverse, ashr

Python (for optional preprocessing and test helpers)

• Python >= 3.10
• packages listed in requirements.txt

Quickstart (example data)

Run an end-to-end smoke analysis:

bash tests/smoke/run_smoke.sh

Expected outputs:

• results/smoke/gene_level_orf_analysis_results_CHLA06.csv
• results/smoke/gene_level_orf_analysis_results_MCF7.csv
• results/smoke/gene_level_orf_analysis_results_all_datasets.csv

Canonical run command

Rscript scripts/ncorf_tiling_analysis.R \
  --config <dataset_config.csv> \
  --mapping <tiling_sgrna_mappings.csv> \
  --gencode <Ribo-seq_ORFs_nochr.bed> \
  --outdir results \
  --num-permutations 10000 \
  --fdr-threshold 0.05 \
  --shrinkage-prior both \
  --seed 42 \
  --make-plots true

Dataset config contract

--config must contain:

• dataset_id
• fitness_path
• lfc_column
• se_column
• optional: mapping_dataset (if using the dataset column in mapping file)

Mapping file contract

--mapping must contain:

• sgrna
• orf
• ensembl_gene_id
• optional: dataset

Fitness file contract

Each fitness CSV must contain:

• sgRNA.Sequence
• one LFC column (named in config)
• one SE/SD column (named in config)