encode-database

What it does

Describes a skill that enables an AI agent to access ENCODE Portal data via REST API. It covers searching for experiments by assay, target, and biosample; retrieving and listing file metadata with download URLs for BED and bigWig files; querying SCREEN for cCREs by genomic region or near a gene; listing biosamples; and querying ChIP-seq experiments for a transcription factor. It also includes example Python code snippets for each task and notes on prerequisites and rate handling. The skill is intended to help annotate variants, identify open chromatin regions, fetch peak files for ChIP/ATAC, and link to RegulomeDB or GWAS databases as needed.

How it works

• Provides a Quick Start and multiple Queries that build HTTP requests to ENCODE endpoints with specific parameters (e.g., search for experiments by assay_title, target.label, biosample_ontology.term_name; retrieve files with type=File, dataset pointing to an experiment; query SCREEN with POST to the SCREEN search endpoint).
• For each operation, the skill specifies parameter keys, endpoint URLs, and typical response handling (e.g., parse @graph, extract fields like accession, biosample_summary, target label, lab title, date_released).
• It includes helper Python functions demonstrating how to perform:
  • Query 1: Experiment search by assay, biosample, target
  • Query 2: Get experiment files and download URLs; download a file
  • Query 3: cCRE region queries via SCREEN API
  • Query 4: Gene-centered cCRE lookup via SCREEN and ENCODE gene search
  • Query 5: List biosamples
  • Query 6: Retrieve all ChIP-seq experiments for a TF
  • Query 7: Retrieve optimal peak sets (partial content shown in excerpt)
• The skill prescribes using the listed commands, file names, and tool names exactly as shown (e.g., requests.get, /search/, ENCODE portals, etc.).

When to use it

• When you need to download TF ChIP-seq peak files (BED) for a specific factor and cell type to annotate regulatory regions.
• When you want ATAC-seq or DNase-seq peaks in a cell type or to identify open chromatin regions near a gene.
• When you need to retrieve SCREEN cCREs overlapping a region or near a gene.
• When building reference regulatory tracks for variant annotation pipelines.
• When exploring which ENCODE experiments are available for a biosample before planning experiments.
• When you need pre-computed regulatory scores (RegulomeDB) or GWAS associations referenced in the description.

What it can touch

• Tools: claude-code (declared tool)
• Endpoints and data surfaces touched include: ENCODE Portal REST API search endpoints, File endpoints, SCREEN API, and ENCODE gene search endpoints. The skill shows usage of URL paths such as /search/, /experiments/{accession}/, and SCREEN endpoints.

Caveats

• License is CC-BY-4.0. No authentication required for public data; for submitter access, an Authorization header may be used.
• Rate handling suggested: time.sleep(0.5) between batch queries to avoid connection resets.
• The examples rely on publicly available ENCODE data and may depend on data availability and endpoint stability.
• The content is code-oriented and includes Python samples for interacting with ENCODE APIs; real use may require adapting to network policies and API changes.