ProteinsGPU8 credits

ProGen2 — Conditional Protein Sequence Generation

Generate novel protein sequences conditioned on family, EC number, or starting motif

ProGen2 (Salesforce, 2023) is a large autoregressive protein language model that can generate novel sequences conditioned on function annotations (family, EC number) or starting motifs. SciRouter exposes ProGen2 via an async API with naturalness scoring.

$0.08

per API call

credits per call

/v1/proteins/generate

API endpoint

Features

Conditional generation by protein family

EC number-conditioned enzyme generation

Starting motif seeding (e.g., 6×His tags)

Naturalness scoring (perplexity-based)

Length range control

Async job submission

Quick Start

ProGen2-xlarge — Python Examplepython

import requests

API_KEY = "sk-sci-your-key-here"

# Generate 5 kinase sequences
resp = requests.post(
    "https://scirouter.ai/v1/proteins/generate",
    json={
        "family": "kinase",
        "ec_number": "2.7.11",
        "length_min": 250,
        "length_max": 350,
        "num_sequences": 5,
    },
    headers={"Authorization": f"Bearer {API_KEY}"}
)
job = resp.json()["data"]
for s in job["result"]["sequences"]:
    print(f"  {s['seq_id']}: len={s['length']} naturalness={s['naturalness']}")