regulomedb-database

What it does

Guides an agent to query the RegulomeDB v2 GET REST API to score variants for regulatory function and to retrieve overlapping evidence such as TF binding, histone marks, DNase peaks, footprints, motifs, eQTLs, and chromatin state. It notes that scores range from 1a (strongest) to 7 (none) and that the skill is intended for GWAS hit prioritization, regulatory variant annotation, and cis-regulatory discovery, with recommendations to use clinvar-database for pathogenicity and gwas-database for trait associations.

How it works

• Use GET /regulome-search/ with regions= to score a single variant (rsID or chr:pos-pos) and receive regulome_score (ranking and probability), along with variants[] and features. The example shows a function regulome_score returning query, ranking, probability, rsids, chrom, and pos.
• For region scans, use GET /regulome-search/ with regions set to a chromosomal window to obtain variants[] and nearby_snps; the skill demonstrates building a DataFrame from the returned variants.
• To parse full evidence, process the @graph array where each entry provides method, target_label, biosample_ontology (term_name, organ_slims, classification), dataset, file, value, chrom, start, end, strand, ancestry, disease_term_name.
• To list TF ChIP-seq hits, filter @graph entries where method == "ChIP-seq" and read target_label and biosample_ontology.
• To obtain tissue-specific regulatory scores, read regulome_score.tissue_specific_scores and rank tissues.
• To find nearby SNPs, inspect nearby_snps for rsid, chrom, coordinates, variation_type, maf.
• The skill provides example code blocks for scoring single variants, region scans, extracting TFs, computing tissue scores, and nearby SNPs.

When to use it

• Prioritizing GWAS hits for regulatory follow-up by identifying regulatory confidence per SNP.
• Annotating a VCF or variant list with regulatory scores to filter for functional relevance.
• Identifying TFs binding near a variant (via ChIP-seq evidence) and reviewing tissue-specific signals.
• Checking whether a non-coding variant overlaps a QTL and active chromatin simultaneously.
• Retrieving all annotated rsIDs in a region for cis-regulatory analysis.

What it can touch

• Uses API endpoints in RegulomeDB (GET /regulome-search/) with regions, genome, and format parameters.
• Returns: regulome_score (ranking, probability, tissue_specific_scores), variants[], features, and @graph evidences; nearby_snps; and optionally @graph details for per-dataset information.

Caveats

• License indicated as CC-BY-4.0; API access is free and requires no authentication. Legacy POST endpoints are non-functional. No published rate limits; suggested to insert time.sleep(0.3) between requests in batch workflows. Ranking is a string (e.g., 1a, 7) and probability is a numeric value; tissue-specific scores map tissues to probabilities. The API can return multiple variants in region queries and may require paging via from parameter.