Access

ImmPrint is the data, and you can reach it however suits your workflow. Every route serves the same versioned snapshot.

R, the immprintr package

The immprintr package delivers ImmPrint as a tidy tibble, in the spirit of msigdbr.

# install.packages("remotes")
remotes::install_github("akmclz/immprintr")

library(immprintr)

ip <- immprintr()                       # latest snapshot as a tidy tibble
immprintr_versions()                    # which snapshots are bundled

# Build a signature for scoring (pan core plus a cell type refinement)
immprintr_signature("IMMPRINT_TGFB_SIGNALLING", "Fibroblast")

# Many signatures as a named list, ready for UCell or AddModuleScore
immprintr_signatures(c("IMMPRINT_IFNG_SIGNALLING", "IMMPRINT_TNF_SIGNALLING"))

Raw data, CSV or JSON

The canonical snapshot is published here for direct, language-agnostic use:

Download CSV Download JSON

Both are also versioned in the repository under data/.

What each column means

The CSV has one row per gene set, cell type and gene membership:

column	meaning
gene_set	ImmPrint set name, e.g. IMMPRINT_IFNG_SIGNALLING
gene_symbol	HGNC gene symbol
pathway_level	signalling tier: 1 = Receptor, 2 = Transducer, 3 = Effector
regulatory_role	the gene’s effect on its own pathway: none (a forward component or output), inhibitory, activating (genuine positive feedback) or context_dependent
cell_type	responding lineage; several lineages are semicolon-delimited (e.g. Epithelial; Fibroblast); pan for the cell type-agnostic core
pmids	semicolon-delimited PubMed ID(s) supporting the membership
reference_gene_set	corroborating MSigDB set, where available

The JSON carries the same fields per record (with pmids as an array and an added pathway_level_label), under a small header giving version, n_gene_sets and n_pairs.

Python

There is no dedicated package yet, so read the published files directly:

import pandas as pd

url = "https://akmclz.github.io/immprint/data/immprint_0.2.0-alpha.csv"
ip = pd.read_csv(url)

# genes of one set, by tier
ip.query("gene_set == 'IMMPRINT_IFNG_SIGNALLING'")[["gene_symbol", "pathway_level"]]

# pan core plus a lineage refinement; cell_type may list several lineages,
# so split on ";" rather than matching the whole string
def has_lineage(ct, want):
    parts = [p.strip() for p in str(ct).split(";")]
    return "pan" in parts or want in parts

tgfb_fibro = ip[(ip.gene_set == "IMMPRINT_TGFB_SIGNALLING")
                & ip.cell_type.apply(has_lineage, want="Fibroblast")]["gene_symbol"]

# a named dict of signatures
sigs = ip.groupby("gene_set")["gene_symbol"].apply(list).to_dict()

Versioning

Every snapshot is immutable. Pin the version your analysis used, both the package immprintr(version = "0.2.0-alpha") and the raw files are version-stamped, so results stay reproducible. See the changelog and how to cite.