DC-TAP-seq paper: analysis code

Code to analyze the K562 and WTC11 DC-TAP-seq CRISPRi screens and to make the figures and tables of:

Ray et al. A systematic survey of distal element-gene regulatory interactions with Direct-Capture Targeted Perturb-seq. bioRxiv 10.1101/2025.09.16.676677

Release used for the paper: v2.0.1 (Zenodo 10.5281/zenodo.23964909).

Data

Data	Where
Raw sequencing and Cell Ranger outputs (pilot, screens, validation screens, CTCF ChIP-seq)	GEO GSE303901
Large intermediate files used by the figures	Zenodo 10.5281/zenodo.23775358 (list: config/large_inputs.txt)
gRNA designs	IGVF Data Portal IGVFFI7272YLVI (K562), IGVFFI0580WJFK (WTC11)
ENCODE annotations	ENCODE Portal; accessions in Supplementary Table 8
Final element-gene results (Table S3)	results/formatted_dc_tap_results/Table_S3_Final_DC_TAP_Seq_Results_w_Chromatin_Categories_on_resized_and_merged_elements_250908_append.tsv; columns described in resources/formatting_dc_tap_results/columns_of_Final_DC_TAP_Seq_Results_table.txt

Repository layout

Path	Contents
workflow/	Snakemake pipeline: Snakefile, rules/*.smk, scripts/, per-rule conda envs in envs/
resources/	Pipeline inputs (design files, published screens, qPCR tables)
results/	Pipeline outputs and figure scripts, one folder per figure
direct_vs_indirect_effects/	Workflow estimating indirect effect rates and the probability of direct effects (Fig. 4, ED Figs. 7, 15, 16)
validations/	Validation screen and CTCF ChIP-seq analyses
crispr_benchmarking/	Inputs for benchmarking E2G models on DC-TAP-seq pairs (Fig. 6)
seqspec/	seqspec templates for DC-TAP-seq libraries
config/	Pipeline config and list of Zenodo inputs

Pipeline steps (workflow/rules/):

1. sceptre_setup.smk: Cell Ranger outputs → SCEPTRE input objects.
2. sceptre_power_analysis.smk: SCEPTRE differential expression and power simulations.
3. create_encode_output.smk: element annotation (promoter, gene body), liftover, ENCODE CRISPR format.
4. formatting_dc_tap_results.smk: confidence intervals, element-gene categories, chromatin categories → Table S3.
5. main_figure_*.smk, figures_wo_positive_controls.smk, supplementary_tables.smk: figures and tables.

System requirements

• OS: Linux x86_64 (tested: CentOS 7, Stanford Sherlock cluster) or macOS arm64 (tested: macOS 26.6, figure rules only).
• Software, installed by pixi from pixi.toml / pixi.lock: Snakemake 7.32.4, Python 3.11, R 4.4.3 with tidyverse, data.table, cowplot, ggExtra, rmarkdown.
• Per-rule software (SCEPTRE, Bioconductor, UCSC liftOver), installed by conda from workflow/envs/*.yml. SCEPTRE comes from the jgalante Anaconda channel, pinned by version, build and sha256:
  • r-sceptre 0.10.1: Katsevich-Lab/sceptre commit 43a4534 (tag v0.10.1)
  • r-sceptre-dev 0.10.1dev (confidence intervals): commit 1517ebc
  • Alternative install: remotes::install_github("Katsevich-Lab/sceptre@<commit>")
• Hardware: figure rules run on a laptop. The full pipeline needs a compute cluster: rules request up to 72 GB RAM and 12 h.

Installation

git clone https://github.com/EngreitzLab/DC_TAP_Paper.git
cd DC_TAP_Paper
git checkout v2.0.1
pixi install        # https://pixi.sh; ~1 min

Per-rule conda environments are built on first use (tens of minutes).

Demo

Make Fig. 4b–d and Extended Data Figs. 7, 15 and 16b,c from files in the repository:

pixi run snakemake --cores 4 \
  results/main_figure_4_and_extdata_figure_7/dc_tapseq_indirect_effects_main_analyses.html \
  results/extended_data_figure_15/dc_tapseq_indirect_effects_extdata_fig15.html \
  results/extended_data_figure_16/3_genome_wide_indirect_effects_extdata_fig16bc.html

Output: PDFs in results/main_figure_4_and_extdata_figure_7/plots/, results/extended_data_figure_15/ and results/extended_data_figure_16/, plus an HTML report for each. Run time: ~20 s on a laptop (Apple M-series, 4 cores).

Reproduce the figures

bash workflow/scripts/download_large_inputs_from_zenodo.sh   # ~0.5 GB from Zenodo, MD5-checked
pixi run main-figures                                         # ~90 min on 8 cores, incl. conda env builds

Figures 4–6 and Extended Data Figs. 7, 15, 16b,c only: pixi run snakemake --cores 4 create_main_figures_4_5_6.

Panel	Rule	Script	Output
1c, 1d	main_figure_1cd_guide_capture_and_moi	workflow/scripts/main_figure_1_and_2/plot_guide_capture_and_moi_knockdown.R	results/main_figure_1_and_2/guide_capture_and_moi_knockdown/
1e	main_figure_1e_umi_saturation	workflow/scripts/main_figure_1_and_2/plot_umi_saturation_dctap_vs_scrna.R	results/main_figure_1_and_2/umi_saturation_dctap_vs_scrna/
2b	distance_by_effect_size_plots_FDR_wo_pos_controls	workflow/scripts/figures_wo_positive_controls/distance_by_effect_size_plots.R	results/figures_wo_positive_controls/{k562,wtc11}_distance_by_es.pdf
2c	effect_size_plots_FDR_wo_pos_controls	workflow/scripts/figures_wo_positive_controls/effect_size_plots.R	results/figures_wo_positive_controls/effect_size_boxplot.pdf
2d	wtc11_multi_moi, make_replicates_plots, main_figure_2d_validation_vs_screen	workflow/scripts/main_figure_1_and_2/, validations/analyses/2026-05-14_compare_low_high_es/compare_es.Rmd	results/main_figure_1_and_2/, validations/results/2026-05-14_compare_low_high_es/
2e	main_figure_2e_dctap_vs_qpcr	workflow/scripts/main_figure_1_and_2/plot_dctap_vs_qpcr_effect_sizes.R	results/main_figure_1_and_2/qpcr_validation/
2f	find_duplicate_pairs	workflow/scripts/main_figure_1_and_2/find_duplicate_pairs.R	results/main_figure_1_and_2/duplicate_pairs_analysis/duplicate_pairs.pdf
2h	slc2a3_effect_size_plot	workflow/scripts/main_figure_1_and_2/slc2a3_effect_size_plot.R	results/main_figure_1_and_2/slc2a3_effect_size_barplot.pdf
2i	main_figure_2i_slc2a3_qpcr	workflow/scripts/main_figure_1_and_2/plot_slc2a3_qpcr.R	results/main_figure_1_and_2/qpcr_validation/
3a, 3b	compare_effect_sizes_to_other_screens_FDR_wo_pos_controls, power_to_detect_change_for_all_datasets_FDR_wo_pos_controls	workflow/scripts/figures_wo_positive_controls/	results/figures_wo_positive_controls/
3c, 5b–d	main_figure_5_chromatin_categories	results/main_figure_5_and_6/chromatin_categories_and_power_estimation.R	results/main_figure_5_and_6/ (5c, 5d: housekeeping_genes/fig5{c,d}_*.pdf)
4b–d, ED 7	main_figure_4_direct_indirect_effects	results/main_figure_4_and_extdata_figure_7/dc_tapseq_indirect_effects_main_analyses.Rmd	results/main_figure_4_and_extdata_figure_7/plots/
6a–d, ED 13	main_figure_6_model_scores	results/main_figure_5_and_6/model_scores.R	results/main_figure_5_and_6/e2g_model_scores/
ED 15	extended_data_figure_15_direct_indirect_rates	results/extended_data_figure_15/dc_tapseq_indirect_effects_extdata_fig15.Rmd	results/extended_data_figure_15/
ED 16	extended_data_figure_16bc_genome_wide_indirect_effects (b,c)	results/extended_data_figure_16/ (steps in its README)	results/extended_data_figure_16/

Derived inputs to Figs. 4–6:

• direct_vs_indirect_effects/compute_direct_vs_indirect_effects/results/: cis/trans hit rates and direct-effect models (made by that folder's Snakemake workflow).
• resources/main_figure_6/expt_pred_merged_annot.txt.gz: output of CRISPR_comparison for DC-TAP-seq pairs.
• results/main_figure_5_and_6/categorized_pairs/: chromatin-categorized CRISPR pairs (chromatin_categories_and_power_estimation.R categorize).

Run the pipeline from Cell Ranger outputs

Download the Cell Ranger outputs from GEO GSE303901 into resources/process_validation_datasets/{K562,WTC11}_DC_TAP_Seq/cell_ranger_output/, then:

snakemake --use-conda --cores <N>

This makes the final results table (results/formatted_dc_tap_results/Final_DC_TAP_Seq_Results_w_Chromatin_Categories_on_resized_and_merged_elements.tsv), the supplementary tables and Figs. 1–2.

To analyze a new DC-TAP-seq screen, build its SCEPTRE inputs (as in sceptre_setup.smk) in results/process_validation_datasets/<sample>/: differential_expression/sceptre_diffex_input.rds, gene_gRNA_group_pairs.rds, gRNA_groups_table.rds, guide_targets.tsv, distances.tsv. Then run differential expression and power simulations:

snakemake --use-conda --cores 4 results/process_validation_datasets/<sample>/differential_expression/results_run_discovery_analysis.rds
snakemake --use-conda --cores 4 results/process_validation_datasets/<sample>/power_analysis/output_0.13gStd_Sceptre_perCRE.tsv

License

MIT; see LICENSE.