Documentation

Rate limits

• Per-key daily quota — see pricing
• Per-IP: 30 inference calls / minute (all plans)
• Per-IP: 5 signups / hour (abuse mitigation)
• Batch endpoints count as 1 quota unit per request, regardless of item count

Docking pipeline integration

Physon's SDF output is docking-ready — MMFF energies + rank + provenance tags are already in the file. Drop straight into AutoDock Vina, Schrödinger Glide, or OpenEye FRED.

End-to-end: SMILES → Physon → Vina

Python example. Assumes you have AutoDock Vina installed (brew install autodock-vina or apt), a receptor prepared as receptor.pdbqt, and a grid box defined.

import httpx, subprocess, tempfile
from rdkit import Chem
from rdkit.Chem import AllChem

API = "https://sharoncs888--physon-mvp-api.modal.run"
KEY = "physon_YOUR_KEY_HERE"
SMILES = "CC(=O)Oc1ccccc1C(=O)O"   # aspirin

# 1. Ask Physon for a docking-ready SDF (ranked by MMFF energy)
r = httpx.post(
    f"{API}/molecular/generate/download",
    headers={"X-API-Key": KEY, "Content-Type": "application/json"},
    json={"smiles": SMILES, "n_conformers": 20, "format": "sdf"},
    timeout=60.0,
)
r.raise_for_status()
sdf_text = r.text

# 2. Convert SDF → PDBQT (Vina input format) via RDKit + Meeko or obabel
with tempfile.NamedTemporaryFile(suffix=".sdf", delete=False) as f:
    f.write(sdf_text.encode())
    sdf_path = f.name

pdbqt_path = sdf_path.replace(".sdf", ".pdbqt")
subprocess.run([
    "obabel", sdf_path, "-O", pdbqt_path, "--gen3d",
], check=True)

# 3. Dock
subprocess.run([
    "vina",
    "--receptor", "receptor.pdbqt",
    "--ligand", pdbqt_path,
    "--config", "vina_config.txt",   # grid box definition
    "--out", "docked.pdbqt",
    "--exhaustiveness", "8",
], check=True)

# 4. Parse Vina scores — best pose is first
with open("docked.pdbqt") as f:
    for line in f:
        if line.startswith("REMARK VINA RESULT:"):
            score = float(line.split()[3])
            print(f"Vina score: {score:.2f} kcal/mol")
            break

Schrödinger Glide (LigPrep input)

Physon SDF drops straight into LigPrep. Skip conformer generation in LigPrep — Physon already produced them.

# 1. Fetch the SDF as above → aspirin.sdf

# 2. Skip LigPrep's conformer step (Physon already generated + ranked):
$SCHRODINGER/ligprep -inp physon.sdf -omae aspirin_prepped.mae \
    -nc 0 \
    -bff 16 \
    -epik    # ionization + tautomerization

# 3. Glide docking (grid + config as usual):
$SCHRODINGER/glide docking.in -HOST localhost -NJOBS 4

Complete workflow — tautomer prep → conformer gen → rank → dock

This is the full enterprise flow. Prepare ionization states, generate conformer ensembles per state, rank by cluster, take medoids into docking.

import httpx

API = "https://sharoncs888--physon-mvp-api.modal.run"
KEY = "physon_YOUR_KEY_HERE"
SMILES = "CC(=O)Oc1ccccc1C(=O)O"

# Step 1 — enumerate protonation states + tautomers
prep = httpx.post(
    f"{API}/molecular/prepare",
    headers={"X-API-Key": KEY},
    json={"smiles": SMILES, "include_tautomers": True, "include_charges": True},
).json()
print(f"{prep['n_states']} docking-ready states")

# Step 2 — rank conformer families per state, keep medoids
all_medoids = []
for state in prep["states"]:
    ranked = httpx.post(
        f"{API}/molecular/rank",
        headers={"X-API-Key": KEY},
        json={"smiles": state["smiles"], "n_conformers": 20,
              "rmsd_threshold": 0.75},
    ).json()
    for cluster in ranked["clusters"]:
        all_medoids.append({
            "smiles": state["smiles"],
            "state_kind": state["kind"],
            "state_weight": state["weight"],
            "cluster_id": cluster["cluster_id"],
            "cluster_size": cluster["size"],
            "medoid_energy_kJ_mol": cluster["medoid_energy_kJ_mol"],
            "medoid_xyz": cluster["medoid_xyz"],
            "medoid_atoms": cluster["medoid_atoms"],
        })

print(f"{len(all_medoids)} unique conformers to dock")
# → feed all_medoids into your docking pipeline
# → weight scores by state_weight for population-corrected affinity

Enterprise plans include curated integrations for KNIME, Pipeline Pilot, and Databricks. Talk to us at /enterprise.