CD39MTK

Reproducible analysis and figure source data for:

Genome-scale CRISPRi identifies an LKB1–PKA–SIK module controlling an ENTPD1-centered T-cell program

This repository connects a genome-scale CRISPRi Perturb-seq screen in primary human CD4⁺ T cells with a 16-gene CD39-related program and rheumatoid arthritis methotrexate treatment data.

Main findings

• Seven knockdowns increased ENTPD1 across resting, 8-hour stimulated and 48-hour stimulated primary human CD4⁺ T cells.
• STK11, PRKAR1A and SIK3 ranked near the top of genome-wide program-level and donor-blocked analyses.
• Whole-transcriptome effects linked the three genes through an LKB1–PKA–SIK–CRTC–CREB pathway.
• ENTPD1 knockdown produced 14 direction-consistent downstream effects.
• The CD39-related program changed during methotrexate exposure in paired rheumatoid arthritis samples.

Repository contents

Path	Contents
manuscript/	Current manuscript and supplementary-information sources
figures/submission_scirep_v4_aggressive/	Eight current main figures in PNG and PDF
figures/supplementary/	Eight supplementary figures used in the submission package
figures/final/	Reusable panel-level and intermediate figures
figures/editable/svg_source_20260728/	Image-free editable SVG figures, grouped layouts, 2 × 4 master canvas, source SVGs and code snapshot
tables/	Analysis-level source tables used by the figures
data/derived/	Three registered derived datasets required for figure generation
metadata/manifests/	Panel-level source-data indexes and checksums
FIGURE_PANEL_SOURCE_MAP.csv	Figure-to-source mapping across all registered panels
scripts/mac/	Local analysis, plotting, assembly and validation scripts
scripts/server/	Server-side scoring and donor-blocked validation scripts
config/	Frozen candidate and readout definitions

The public repository contains analysis-level results. It excludes raw sequencing matrices, participant-level clinical matrices, full server checkpoints, caches, credentials, downloaded reference genomes and restricted gene-set files.

Quick start

Install uv and create the locked environment:

uv sync --extra figures --frozen

Verify every released file:

sha256sum -c MANIFEST.sha256

On macOS, use shasum -a 256 -c MANIFEST.sha256.

Rebuild the analytical figure layers:

uv run python scripts/mac/64_build_manuscript_figures.py
uv run python scripts/mac/65_build_extended_figures.py
uv run python scripts/mac/66_build_graphical_abstract.py
uv run python scripts/mac/67_build_design_and_selection_figures.py
uv run python scripts/mac/70_score_continuous_panel_program.py
uv run python scripts/mac/76_build_supplementary_figures.py
uv run python scripts/mac/84_build_aggressive_hybrid_figures.py

Rebuild the donor-blocked vector panel from compact, plot-ready tables and assemble the editable SVG release:

uv run python scripts/mac/90_build_donor_blocked_svg_from_tables.py
uv run python scripts/mac/89_build_true_vector_svg_package.py

The editable package contains eight individual figures, recommended Figure 3 + 4 and Figure 5 + 6 combinations, and SCIREP_v4_Figures1-8_MASTER_2x4_TRUE_VECTOR.svg. Its manifest records 20 SVG files; every file is XML-parseable, retains text as <text>, and contains no <image> or <feImage> elements.

The locked composition scripts verify exact input-image hashes:

uv run python scripts/mac/71_build_figure2_submission_composite.py
uv run python scripts/mac/72_validate_figure2_submission_composite.py
uv run python scripts/mac/74_build_main_figure_composites_and_source_index.py
uv run python scripts/mac/75_validate_main_figures_source_data.py

PNG and PDF hashes can vary when fonts or rendering libraries differ. The released source tables and numerical validations define the scientific result; the bundled images define the locked visual reference.

Source data and provenance

FIGURE_PANEL_SOURCE_MAP.csv records the source path, filters, fields, grain, generator and checksum for each figure view. DATA_SOURCES.md lists upstream datasets, accessions and reuse terms.

All clinical tables in this release are gene-level or model-level summaries. They contain no names, contact details, payment information or direct participant identifiers.

Citation

The exact v0.2.1 archive is available at doi:10.5281/zenodo.21636585. The concept DOI doi:10.5281/zenodo.21636584 always resolves to the latest archived version. Citation metadata are also provided in CITATION.cff. Please cite the primary Perturb-seq, rheumatoid arthritis cohort and database publications listed in the manuscript and DATA_SOURCES.md.

License

• Original software in this repository: MIT License.
• Original manuscript text, figures and contributor-created documentation: CC BY 4.0.
• Derived data tables: see DATA_LICENSE.md and DATA_SOURCES.md; upstream terms and attribution requirements remain in force.