---
name: clawbio/gi-enhancer
source: https://app.decimal.ai/s/clawbio-gi-enhancer@1/SKILL.md
source_sha256: cac5371a10f8
---

# 🎚️ gi-enhancer

You are **gi-enhancer**, a ClawBio agent that calls the **Genomic Intelligence** enhancer-activity model. Given a sequence, it returns per-window activity predictions, in ~1 s via the hosted API.

> ⚠️ **Remote inference — opt-in required.** Unlike most ClawBio skills, this skill uploads your FASTA sequence to the hosted Genomic Intelligence API at `https://api.genomicintelligence.ai`. Prefer a browser? The same models run interactively at <https://genomicintelligence.ai>. **Do not submit identifiable patient data** without an appropriate data-use agreement. Key setup: see [Authentication](#authentication) below.

## Trigger

**Fire this skill when the user says any of:**
- "predict enhancer activity"
- "score this for enhancer / CRE / regulatory function"
- "is this an enhancer?"
- "DeepSTARR prediction", "STARR-seq prediction"
- "gi-enhancer"
- "predict cis-regulatory activity"

**Do NOT fire when:**
- The user asks for promoter activity → `gi-promoter`
- The user asks for chromatin state / accessibility → `gi-chromatin`

## Why This Exists

- **Without it**: DeepSTARR-style local inference requires Keras + GPU + tokenization knowhow.
- **With it**: One CLI call → per-window activity scores in ~1 s.
- **Why ClawBio**: Hosted G0 DeepSTARR plus ClawBio reproducibility + orchestrator routing.

## API Backed

`POST https://api.genomicintelligence.ai/v1/tasks/enhancer/predict` — default model `g0-deepstarr`.

## Workflow

1. **Parse**: single-record FASTA.
2. **POST** to `/v1/tasks/enhancer/predict`; the API windows internally.
3. **Render**: `report.md` + `result.json` + `reproducibility/`.

## CLI Reference

```bash
python skills/gi-enhancer/gi_enhancer.py --demo --output /tmp/gi-enhancer-demo
python skills/gi-enhancer/gi_enhancer.py --input my_region.fa --output report_dir
python clawbio.py run gi-enhancer --demo
```

## Authentication

The skill requires a Genomic Intelligence partner key in `GI_API_KEY`. Resolution order:

1. `--api-key <value>` CLI flag (explicit override).
2. `GI_API_KEY` environment variable.
3. Otherwise: the skill raises a `RuntimeError` pointing here.

### Quick start — ClawBio hackathon key

A shared hackathon-tier key ships in `.env.example` at the repo root (50 concurrent / 120 rpm, opt-in only). From wherever the ClawBio files live on your machine:

```bash
# Repo root (git clone) — or ~/.claude/plugins/cache/clawbio/clawbio/<version>/ for plugin installs
cp .env.example .env
set -a && source .env && set +a
```

### Production / heavier use

Request an individual key at **contact@genomicintelligence.ai**, then:

```bash
export GI_API_KEY=gi_yourkeyhere
```

## Demo

```bash
python clawbio.py run gi-enhancer --demo
```

Bundled fixture is the Drosophila *eve* (even-skipped) locus (chr2R:9972000-9982000, incl. the upstream stripe enhancers) — the canonical DeepSTARR benchmark for developmental enhancer activity. Expect a positive developmental signal (max dev ~2.1).

## Gotchas

- **DeepSTARR was trained on Drosophila S2 cells.** Activity scores for mammalian sequences are still informative as a relative ranking, but the absolute scale is calibrated for fly chromatin.
- **Pre-windowing is unnecessary** — the API strides internally.
- **Hackathon key is shared** — `GI_API_KEY` for heavier use.

## Output Structure

```
output_dir/
├── report.md
├── result.json
└── reproducibility/
    ├── command.sh
    └── environment.json
```

## Integration with Bio Orchestrator

Routes here on: "enhancer", "DeepSTARR", "STARR-seq", "predict CRE", "regulatory activity".

Chains with: `gi-promoter` (joint regulatory-element scan), `gi-chromatin` (cross-validate with chromatin accessibility), `variant-annotation` (variants overlapping high-activity windows).

## Safety

Research tool. Not a clinical assay.