cellink.tl.external.genescores_to_annotations

cellink.tl.external.genescores_to_annotations(genescores, roadmap_links, abc_links, gene_annotation, bim_prefix, out_dir, *, tissue='BLD', chromosomes=None, window_kb=100, save_bedgraphs=True, use_bedtools=True, **link_kwargs)

Full Step 2: gene scores → bedgraphs → per-chromosome SNP annotations.

Produces TWO annotation strategies per program, matching the sc-linker paper: - ABC_Road_{tissue}: Roadmap ∪ ABC enhancer-gene links (weighted) - 100kb: gene body ± window_kb (unweighted)

Parameters:

• genescores (DataFrame) – DataFrame (genes X programs) with probabilistic scores in [0, 1].

• roadmap_links (DataFrame) – Roadmap enhancer-gene links (from load_roadmap_links).

• abc_links (DataFrame) – ABC enhancer-gene links (from load_abc_links).

• gene_annotation (DataFrame) – Gene coordinate file (from load_gene_annotation).

• bim_prefix (str) – Prefix for per-chromosome BIM files. Example: "refs/1000G.EUR.QC." → files ...1.bim, ...2.bim, …

• out_dir (str | Path) – Root output directory. Creates sub-dirs per program and strategy.

• tissue (str (default: 'BLD')) – Tissue code (used only for naming the strategy folder).

• chromosomes (list[int] | None (default: None)) – Chromosomes to process.

• window_kb (int (default: 100)) – Window for 100kb strategy.

• save_bedgraphs (bool (default: True)) – Whether to save bedgraph files alongside annotation files.

• use_bedtools (bool (default: True)) – Use bedtools binary for interval merging and annotation.

Return type:

dict[str, dict[str, str]]

Returns:

dict {program_name: {"ABC_Road": annot_prefix, "100kb": annot_prefix}} where annot_prefix is a string like "out_dir/program/ABC_Road_BLD/program." (without the chromosome number, to be passed to LDSC).