biodiscovery 0.3.0

Unofficial Python SDK for Amazon Bio Discovery — AI-powered antibody design

biodiscovery-python

The unofficial Python SDK for Amazon Bio Discovery.

Design antibodies with AI. Programmatically.

pip install biodiscovery

⚠️ Unofficial. This package is not affiliated with or endorsed by Amazon Web Services. Built by Nodes Bio, Inc. — AI Synthesis for Science and Medicine.

---

Quick Start

from biodiscovery import BioDiscoveryClient

client = BioDiscoveryClient(
    application_id="your-app-id",
    access_key_id="AKIA...",
    secret_access_key="...",
    session_token="...",
)

# Get a project
project = client.get_project("f29a1f5a33504ca885d0bea53d3570ba")
print(project.name, project.status)

# Get experiment results
experiment = client.get_experiment(
    project_id=project.project_id,
    experiment_id="f4ef6a1e01c84f4ea6bf29bf2028e5e5",
)
print(f"Cost: {experiment.estimated_experiment_units} EU")
print(f"Status: {experiment.status}")

# Inspect a module's IO spec
module = client.get_module("rfantibody")
print(module.description)
print(module.io_spec.inputs)

# Get a recipe DAG
recipe = client.get_recipe("8f2e9b3c7d154aa6b2c8f1e9d0a3b5c4")
for step in recipe.steps:
    print(f"  {step.step_name} → {step.module_id}")

# List wet lab orders
orders = client.list_wetlab_orders(project.project_id, experiment.experiment_id)

Authentication

Amazon Bio Discovery uses AWS SigV4 signing with temporary credentials obtained via Identity Center SSO. The SDK handles signing automatically — you provide the credentials from your Bio Discovery session.

To get your credentials:

1. Log in to biodiscovery.aws.com
2. Open browser DevTools → Network tab
3. Find the /getcredentials response
4. Copy applicationId, credentials.accessKeyId, credentials.secretAccessKey, credentials.sessionToken

API Coverage (v0.2.0 — 55 methods)

Projects

list_projects · get_project · create_project · update_project · delete_project

Experiments

list_experiments · get_experiment · create_experiment · update_experiment · delete_experiment · start_experiment · predict_experiment_configuration

Results

list_results · get_result_data · get_result_metadata

Jobs

list_jobs · get_job · create_job · start_job · update_job · delete_job

Modules & Recipes

list_hosted_modules · get_module · list_hosted_recipes · get_recipe · list_custom_recipes · get_custom_recipe · create_custom_recipe · update_custom_recipe · delete_custom_recipe · start_custom_recipe_publish

Wet Lab (CRO Integration)

list_wetlab_providers · list_wetlab_orders · get_wetlab_order · get_wetlab_order_estimates · create_wetlab_order · cancel_wetlab_order · delete_wetlab_order · export_wetlab_order_results

Datasets & Files

list_datasets · get_dataset_metadata · import_dataset · export_dataset · update_dataset · start_dataset_multipart_upload · resume_dataset_multipart_upload · complete_dataset_multipart_upload · get_file_metadata · export_file · delete_file

Sharing & Permissions

list_resource_permissions · create_resource_permission · update_resource_permission · delete_resource_permission · update_principal_permissions · list_principals

Identity Center

list_idc_users · list_idc_groups · associate_idc_principal · disassociate_idc_principal

Account

get_user_details · list_feature_flags · get_usage

Module Catalog

The SDK includes a reference catalog of known Bio Discovery modules:

from biodiscovery.catalog import MODULES, RECIPES, CRO_PARTNERS

# All 40+ hosted bioFMs
for module_id, description in MODULES.items():
    print(f"{module_id}: {description}")

# Known recipe IDs
print(RECIPES["de_novo_design"])

# CRO partners
for partner, info in CRO_PARTNERS.items():
    print(f"{partner}: {info}")

Models

All API responses are parsed into typed Pydantic models:

• Project — project metadata and status
• Experiment — experiment config, parameters, EU cost, timing
• Module — bioFM module with parsed IO specification
• Recipe — workflow DAG with steps and dependency piping
• WetLabOrder — CRO order status and assay details

Before You Design: Understand the Biology

Amazon Bio Discovery designs antibodies. But which target should you design against?

Jarvis sends your biology question to five AI models simultaneously and returns a consensus answer with a confidence score. Use it to evaluate targets, review literature, and build confidence before you spend Experiment Units.

MedMap turns plain-English biology questions into interactive pathway maps. Visualize the disease mechanism, identify druggable nodes, then take those targets into Bio Discovery.

Research → Visualize → Design → Test. That's the workflow.

Contributing

Contributions welcome:

• Async client (httpx.AsyncClient)
• CLI tool
• Better credential management (browser cookie extraction, SSO flow)
• Response model coverage for newer endpoints
• Integration examples and notebooks

License

Apache 2.0 — see LICENSE.

---

Built by Nodes Bio, Inc. — AI Synthesis for Science and Medicine.