Closed-Loop CLI

mshale-loop v0.1.0 — the prediction-experiment-feedback CLI. Queries mshale-api for top protocol recommendations, manages multi-round optimization campaigns, and triggers model retraining when new outcome data arrives. State is stored locally at ~/.mshale/campaigns.json.

Loop Iteration

Quickstart

pip install mshale-loop

export MSHALE_API_BASE=https://api.mishale.bio
export MSHALE_API_KEY=msk_...

# Start a new optimization campaign
mshale-loop campaign create \
  --target "fibroblast → neuron" \
  --species human \
  --budget 10

# Get top 3 protocol recommendations
mshale-loop recommend \
  --campaign my-campaign \
  --top 3

# Submit a wet-lab result
mshale-loop submit \
  --campaign my-campaign \
  --spec protocol_rank1.json \
  --efficiency 0.23 \
  --metric MAP2_positive \
  --confidence high

# Check campaign status
mshale-loop campaign show my-campaign

# View C-index improvement history
mshale-loop history --campaign my-campaign

CLI Reference

Python API

from mshale_loop.campaign import create_campaign, get_campaign, record_result
from mshale_loop.client import MshaleClient

client = MshaleClient(
    api_base="https://api.mishale.bio",
    api_key="msk_...",
)

# Create campaign (stored at ~/.mshale/campaigns.json)
campaign = create_campaign(
    name="fibroblast-neuron-q1",
    target="fibroblast → neuron",
    species="human",
    budget=10,
)

# Get recommendations
recs = client.recommend(campaign_id=campaign.id, top_n=3)
for r in recs:
    print(r.protocol_id, r.predicted_score)

# Record a wet-lab result
updated = record_result(
    campaign_id=campaign.id,
    protocol_id="prot_abc123",
    efficiency=0.23,
    metric="MAP2_positive",
)
print(f"Best efficiency so far: {updated.best_efficiency:.2f}")

Convergence Monitoring