GoldBEAM pilot program

Request access to GoldBEAM

GoldBEAM is open to a small number of research groups during our pilot. Pilot access is free for a trial period. Tell us about your research and we will set up your account.

Already approved?

GoldBEAM pilot program

Checking your account

One moment while we load your access details.

Email contact@swaev.com
GoldBEAM is in pilot. If something looks wrong or could work better, send us a report. Your feedback directly shapes the platform.

Pilot trial

Usage

Free pilot trial

Megabase Usage (cumulative) 0.00 / 50.00 MB
0.0%
Access Pilot
Concurrency 2 Active Slots
Active Jobs 0 Running

Terminal Credentials

GoldBEAM API Key

Paste this key into the GoldBEAM TUI when it asks for your API key. Keep it private: anyone with the key can use your allowance.

••••••••••••••••••••••••••••••••

Active Compute Slots

Concurrent Tasks Management

Predictions you have queued or running. Each one uses one of your concurrent job slots; cancel a job here to free its slot.

No concurrent tasks currently active.

Contact map viewer

Measured Hi-C (example)

Until you run a prediction, this viewer shows one window of real measured Hi-C from a public benchmark, so you can see what experimental data looks like at the model's resolution. When a prediction completes, it replaces the example here. Domain boundaries are called in your browser from the insulation score of whatever map is shown.

Research preview. Human (hg38) sequence only. Input: exactly 1,048,576 bp. Output: a 448 × 448 contact map at 2,048 bp resolution, covering the central 917,504 bp of your sequence.
Position Hover the map
Value –

Loading example…

Scroll to zoom · drag to pan · dbl-click to reset

Insulation score ▼ boundaries
Palette

Live Inference

Submit a Prediction

Paste exactly 1,048,576 bp of human (hg38) sequence in FASTA or raw format. GoldBEAM tokenizes and submits the job using your API key, and the predicted contact map appears in the viewer above when complete.

Human (hg38)

Session

Prediction History

Terminal Client

Install the GoldBEAM TUI

One command installs the full interactive TUI → chromatin structure prediction, TAD browser, motif deletion sweeps, saliency maps, and publication-ready exports. Runs entirely in your terminal.

  • Requires Python 3.9+ and an active API key
  • bash · zsh · fish → shell auto-detected
  • Adds swaev to your PATH automatically
One-command install 
# Downloads client, installs deps, adds swaev to PATH
curl -fsSL https://raw.githubusercontent.com/CK5515/SWAEV_TUI/main/install.sh | bash
Launch 
# Start the bacteriophage TUI
swaev
Uninstall 
# Removes ~/.swaev/ and the swaev command
rm -rf ~/.swaev ~/.local/bin/swaev && echo "SWAEV TUI uninstalled."

Developer Reference

API Quick-Start

Every example below is copy-paste ready. Your API key is auto-filled from the Key Management section above. Tokenisation convention: A=0 C=1 G=2 T=3 N=4.

1
Export your API key
Set it once in your shell and every command below picks it up automatically.
Shell · export
export SWAEV_API_KEY="your_api_key"
2
Confirm the gateway is reachable
No key required. Returns {"status":"ok"} when the service is live.
curl · GET /healthz
curl https://gateway.swaev.com/healthz
3
Submit a prediction
POST a tokenised integer array of exactly 1,048,576 bp. Returns a job_id immediately, and inference runs asynchronously.
curl · POST /v1/predict
# Replace [...] with your tokenised sequence (800k–1.2M integers)
curl -X POST "https://gateway.swaev.com/v1/predict" \
  -H "X-API-Key: $SWAEV_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"sequence": [0,1,2,3,0,1,2,3,0,2,1,3,2,0,1,3,0,2]}' 
# → {"job_id": "3f7a...", "status": "queued"}
4
Poll for the 256×256 contact map
Swap in your job_id from step 3. Status cycles queued → processing → completed. The matrix field is a 256×256 float array.
curl · GET /v1/results/:job_id
curl "https://gateway.swaev.com/v1/results/YOUR_JOB_ID" \
  -H "X-API-Key: $SWAEV_API_KEY"
# → {"status": "completed", "matrix": [[0.82, 0.61, …], …]}
5
Python: tokenise a FASTA file and submit
Reads any FASTA file, converts bases to GoldBEAM tokens, and queues a prediction. Run pip install requests first.
Python · submit
import requests

GATEWAY = "https://gateway.swaev.com"
API_KEY = "your_api_key"
TOKEN   = {"A": 0, "C": 1, "G": 2, "T": 3}   # N → 4

# Tokenise → drop header lines, upper-case, map bases
with open("region.fa") as f:
    seq = "".join(l.strip() for l in f if not l.startswith(">"))
tokens = [TOKEN.get(b.upper(), 4) for b in seq]  # exactly 1,048,576 bp

r = requests.post(f"{GATEWAY}/v1/predict",
    headers={"X-API-Key": API_KEY},
    json={"sequence": tokens}, timeout=60)
r.raise_for_status()
job_id = r.json()["job_id"]
print(f"Queued: {job_id}")
6
Python: full pipeline → tokenise → submit → poll → save NumPy
End-to-end script. Blocks until inference completes, then saves the 256×256 contact matrix as a .npy file ready for matplotlib, cooler, or any downstream tool.
Python · full pipeline
import requests, time, numpy as np

GATEWAY = "https://gateway.swaev.com"
API_KEY = "your_api_key"
TOKEN   = {"A": 0, "C": 1, "G": 2, "T": 3}

# 1. Tokenise FASTA (exactly 1,048,576 bp)
with open("region.fa") as f:
    seq = "".join(l.strip() for l in f if not l.startswith(">"))
tokens = [TOKEN.get(b.upper(), 4) for b in seq]

# 2. Submit
r = requests.post(f"{GATEWAY}/v1/predict",
    headers={"X-API-Key": API_KEY},
    json={"sequence": tokens}, timeout=60)
r.raise_for_status()
job_id = r.json()["job_id"]
print(f"Queued: {job_id}")

# 3. Poll until complete
while True:
    r = requests.get(f"{GATEWAY}/v1/results/{job_id}",
                     headers={"X-API-Key": API_KEY})
    data = r.json()
    if data["status"] == "completed":
        matrix = np.array(data["matrix"])      # shape (256, 256)
        np.save("contact_map.npy", matrix)
        print(f"✓ Saved contact_map.npy  shape={matrix.shape}  max={matrix.max():.3f}")
        break
    if data["status"] == "failed":
        raise RuntimeError("Inference failed: check the sequence is exactly 1,048,576 bp.")
    time.sleep(3)