tryscout-sdk 0.1.0a3

Python SDK for Scout biomedical workflow APIs.

TryScout SDK

Python SDK for building biomedical workflows on top of the Scout API.

Install

pip install tryscout-sdk

Quickstart

import scout
from scout.tools import Target, chem, docking
from scout.workflow import Storage, pipeline, rank, report

scout.initialize(api_key="YOUR_SCOUT_API_KEY", base_url="https://api.tryscout.dev")

target = Target.resolve("EGFR")

pipe = pipeline("hit_triage")
pipe.configure(storage=Storage.local(root=".scout"))
pipe.input("compounds")
pipe.input("protein")

pipe.step("deduplicate", lambda compounds: chem.deduplicate(compounds))

pipe.step("filter", lambda deduplicate: deduplicate)

diffdock = pipe.branch("diffdock", from_step="filter")
diffdock.step(
    "dock",
    lambda filter, protein: docking.predict(
        protein=protein,
        ligands=filter,
    ),
)
diffdock.step(
    "rank",
    lambda filter, dock: rank.by_weighted_score(
        items=list(filter),
        features={"dock": dock.to_scores()},
        weights={"dock": 1.0},
    ),
)

vina = pipe.branch("vina", from_step="filter")
vina.step(
    "dock",
    lambda filter, protein: docking.predict(
        protein=protein,
        ligands=filter,
    ),
)
vina.step(
    "rank",
    lambda filter, dock: rank.by_weighted_score(
        items=list(filter),
        features={"dock": dock.to_scores()},
        weights={"dock": 1.0},
    ),
)

job = pipe.run(
    branches=["diffdock", "vina"],
    compounds="hits.csv",
    protein="egfr.pdb",
)

print(job.id)
print(job.status())
print(job.progress())
print(job.current_step())

run = job.wait()
report.export(run.output("diffdock.rank"), "ranked_hits.html")

Tool calls remain synchronous by default:

pred = property_prediction.predict(molecules="hits.csv")

Optional asynchronous submission is available with .submit(...):

job = docking.submit(protein="egfr.pdb", ligands=["CCO", "CCN"])
result = job.wait()

Pipeline jobs support controls:

job.cancel()
resumed = job.resume()
retried = job.retry(step="diffdock.dock")
branched = job.branch(from_step="filter", name="vina")

# unified resume control on run():
resumed = pipe.run(compounds="hits.csv", protein="egfr.pdb", resume="auto")
from_dock = pipe.run(compounds="hits.csv", protein="egfr.pdb", resume="from:diffdock.dock")