tooluniverse-regulatory-genomics

Annotate regulatory elements and predict non-coding variant effects using JASPAR, ENCODE, RegulomeDB, and deep-learning models.

1.7k|254|Updated Mar 3, 2025
One-click install
npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-regulatory-genomics
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tooluniverse-regulatory-genomics
Source: https://github.com/mims-harvard/ToolUniverse/tree/main/plugins/tooluniverse/skills/tooluniverse-regulatory-genomics
Command: npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-regulatory-genomics

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Determining what regulates a gene or whether a non-coding variant is functionally important requires combining motif databases, experimental ChIP-seq data, chromatin annotations, and variant scoring — evidence scattered across JASPAR, ENCODE, UCSC, and RegulomeDB that is tedious to query and integrate manually.

Core Features & Use Cases

  • TF motif and binding-site discovery: Retrieve TF binding motifs from JASPAR and curated direct TF-DNA binding sites from UniBind, then cross-reference with ENCODE ChIP-seq experiments.
  • Regulatory element annotation: Map cCREs (promoters, enhancers, CTCF sites) around any genomic region via UCSC and ENCODE annotations, ChromHMM states, and ATAC-seq/DNase-seq data.
  • Variant impact assessment: Score regulatory variants with RegulomeDB rankings, and predict quantitative effects of non-coding variants from raw sequence using AlphaGenome, Enformer, Borzoi, ChromBPNet, or Evo 2.
  • Use Case: Given a GWAS hit like rs1234567, score its regulatory impact with RegulomeDB, check whether it overlaps a cCRE, find which TF motifs it disrupts, and predict its effect size on expression with a sequence-based model.

Quick Start

Ask the agent to determine whether a specific SNP overlaps a regulatory element and predict its effect on gene expression using ENCODE, RegulomeDB, and sequence-based models.

Frequently Asked Questions about tooluniverse-regulatory-genomics

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find transcription factor binding sites near a gene?

Search JASPAR with jaspar_search_matrices using the TF name to get its binding motif, then query ENCODE_search_experiments with assay_title "TF ChIP-seq" and the TF target to find experimental binding data. UniBind provides curated direct binding sites with downloadable BED files.

How to check if a SNP affects a regulatory element?

Query RegulomeDB_query_variant with the rsID to get a regulatory ranking from 1a (strongest evidence) to 7 (no evidence). Then call UCSC_get_encode_cCREs with the variant's GRCh38 coordinates to check overlap with annotated enhancers or promoters.

What is the difference between RegulomeDB and Enformer for variant scoring?

RegulomeDB ranks variants by overlap with existing annotations like eQTLs, TF binding, and DNase peaks, but only works for known rsIDs. Enformer and similar models predict quantitative effect sizes directly from sequence, so they can score novel variants with no prior annotation.

Does UCSC cCRE annotation support hg19 coordinates?

No, UCSC_get_encode_cCREs requires GRCh38 (hg38) coordinates with chromosome format like "chr1". Variants on hg19 must be lifted over to GRCh38 before querying.

Why does ENCODE_search_experiments return no results?

The assay_title parameter must match ENCODE's controlled vocabulary exactly, such as "TF ChIP-seq" rather than the generic "ChIP-seq". If filters are too restrictive, remove biosample_term_name and retry with broader criteria.

Can I predict variant effects without an AlphaGenome API key?

Yes, Enformer, Borzoi, and ChromBPNet run through a remote MCP server without a hosted API key, and Evo 2 uses an NVIDIA API key. If no model access is provisioned, fall back to annotation-based tools like RegulomeDB and ENCODE.