rowan

rowan

Overview

Rowan is a cloud quantum chemistry platform that exposes DFT and semiempirical calculations through a Python SDK (rowan). Submit calculations (geometry optimization, conformer generation, torsional scans, single-point energies) from Python scripts or Jupyter notebooks, and retrieve results — energies, geometries, partial charges, frontier orbital energies — without managing Gaussian, ORCA, or Psi4 installations. Rowan handles job queuing, execution, and storage. A free tier is available for academic and exploratory use.

When to Use

• Geometry optimization of small molecules: Getting accurate equilibrium geometries for drug candidates, fragments, or building blocks using DFT.
• Conformer generation with energy ranking: Generating and optimizing multiple conformers to identify the lowest-energy conformation for docking or property prediction.
• Torsional potential scans: Mapping the energy profile along a rotatable bond to understand conformational preferences.
• Quantum mechanical property calculation: Computing dipole moments, partial charges (Mulliken, ESP), HOMO/LUMO energies, and electrostatic potential surfaces.
• Energy minimization before docking: Refining ligand geometries before input to structure-based docking tools (DiffDock, AutoDock Vina).
• Comparing isomer stability: Calculating relative energies of tautomers, stereoisomers, or constitutional isomers.
• For large-scale conformer screening (>1000 molecules), use RDKit's ETKDGv3 + MMFF (force field level, no cloud cost).
• For protein-scale quantum mechanics/molecular mechanics (QM/MM), specialized packages like ORCA + CP2K are needed.

Prerequisites

• Python packages: rowan (official Python SDK)
• Account: Free account at https://rowan.chem.ucla.edu/ (academic) or https://rowanquantum.com/
• API key: Set ROWAN_API_KEY environment variable after account creation
• Data requirements: Molecular structures as SMILES strings or XYZ coordinate blocks

pip install rowan

# Set API key (add to .bashrc or .env)
export ROWAN_API_KEY="your_api_key_here"

Quick Start

import rowan

# Authenticate (uses ROWAN_API_KEY environment variable automatically)
client = rowan.RowanClient()

# Run geometry optimization of aspirin at GFN2-xTB level
job = client.compute(
    smiles="CC(=O)Oc1ccccc1C(=O)O",
    method="gfn2-xtb",
    tasks=["optimize"],
)
print(f"Job ID: {job.id}, Status: {job.status}")

# Wait for completion and retrieve energy
result = client.wait(job.id)
print(f"Energy: {result.energy:.6f} Hartree")
print(f"Optimized geometry atoms: {len(result.geometry.atoms)}")

Core API

Module 1: Client Initialization and Authentication

import rowan
import os

# Option 1: automatic (reads ROWAN_API_KEY env variable)
client = rowan.RowanClient()

# Option 2: explicit key
client = rowan.RowanClient(api_key=os.environ["ROWAN_API_KEY"])

print(f"Authenticated as: {client.user.email}")
print(f"Organization: {client.user.organization}")

Module 2: Geometry Optimization

Optimize a molecular geometry to the nearest local minimum.

import rowan

client = rowan.RowanClient()

# GFN2-xTB semiempirical (fast, good for conformer screening)
job_xtb = client.compute(
    smiles="CCc1ccc(cc1)NC(=O)C",    # paracetamol
    method="gfn2-xtb",
    tasks=["optimize"],
)
result_xtb = client.wait(job_xtb.id)
print(f"xTB optimized energy: {result_xtb.energy:.6f} Hartree")
print(f"Geometry: {len(result_xtb.geometry.atoms)} atoms")

# DFT optimization: B3LYP/6-31G* (accurate, slower)
job_dft = client.compute(
    smiles="CCc1ccc(cc1)NC(=O)C",
    method="b3lyp",
    basis_set="6-31g*",
    tasks=["optimize"],
    solvent="water",               # implicit solvent (SMD model)
)
result_dft = client.wait(job_dft.id)
print(f"B3LYP/6-31G* energy (water): {result_dft.energy:.6f} Hartree")
print(f"Dipole moment: {result_dft.dipole_moment:.3f} Debye")

Module 3: Conformer Generation

Generate multiple 3D conformers and rank by energy.

import rowan
import pandas as pd

client = rowan.RowanClient()

# Generate and optimize 10 conformers at GFN2-xTB level
job = client.compute(
    smiles="CC(C)CC1=CC=C(C=C1)C(C)C(=O)O",  # ibuprofen
    method="gfn2-xtb",
    tasks=["conformers"],
    n_conformers=10,
)
result = client.wait(job.id)

# Rank conformers by relative energy
conformers = result.conformers
df = pd.DataFrame([
    {"conformer_id": i,
     "energy_hartree":    c.energy,
     "rel_energy_kcal":   (c.energy - min(c2.energy for c2 in conformers)) * 627.509}
    for i, c in enumerate(conformers)
]).sort_values("rel_energy_kcal")

print(df.to_string(index=False))
print(f"\nLowest energy conformer ID: {df.iloc[0]['conformer_id']}")

Module 4: Torsional Scan

Map energy as a function of a dihedral angle.

import rowan
import numpy as np
import matplotlib.pyplot as plt

client = rowan.RowanClient()

# Scan the C-C=C-C dihedral of butene
job = client.compute(
    smiles="CC=CC",             # but-2-ene
    method="gfn2-xtb",
    tasks=["torsion_scan"],
    torsion_atoms=[0, 1, 2, 3],  # atom indices defining dihedral
    n_scan_points=36,             # 36 points = 10-degree steps
)
result = client.wait(job.id)

angles  = [point.angle  for point in result.torsion_scan]
energies = [point.energy for point in result.torsion_scan]
rel_e = [(e - min(energies)) * 627.509 for e in energies]  # convert to kcal/mol

fig, ax = plt.subplots(figsize=(7, 4))
ax.plot(angles, rel_e, "o-", color="steelblue")
ax.set_xlabel("Dihedral angle (degrees)")
ax.set_ylabel("Relative energy (kcal/mol)")
ax.set_title("Torsional potential: but-2-ene C-C=C-C dihedral")
plt.tight_layout()
plt.savefig("torsion_scan.png", dpi=150)
print("Torsion scan saved -> torsion_scan.png")

Module 5: Single-Point Properties

Compute electronic properties at a fixed geometry (HOMO/LUMO, partial charges, ESP).

import rowan

client = rowan.RowanClient()

# Single-point DFT: get frontier orbital energies and partial charges
job = client.compute(
    smiles="c1ccccc1N",          # aniline
    method="b3lyp",
    basis_set="6-311+g**",
    tasks=["single_point", "partial_charges", "orbitals"],
)
result = client.wait(job.id)

print(f"HOMO energy: {result.homo_energy:.4f} eV")
print(f"LUMO energy: {result.lumo_energy:.4f} eV")
print(f"HOMO-LUMO gap: {result.lumo_energy - result.homo_energy:.4f} eV")
print(f"Dipole moment: {result.dipole_moment:.3f} Debye")

# Partial charges (Mulliken)
for i, (atom, charge) in enumerate(zip(result.geometry.atoms, result.partial_charges)):
    print(f"  Atom {i} ({atom.symbol}): {charge:+.4f} e")

Module 6: Batch Job Submission

Submit multiple molecules concurrently.

import rowan
import pandas as pd
from concurrent.futures import ThreadPoolExecutor, as_completed

client = rowan.RowanClient()

smiles_list = [
    ("aspirin",      "CC(=O)Oc1ccccc1C(=O)O"),
    ("caffeine",     "Cn1cnc2c1c(=O)n(c(=O)n2C)C"),
    ("paracetamol",  "CC(=O)Nc1ccc(O)cc1"),
    ("ibuprofen",    "CC(C)Cc1ccc(cc1)C(C)C(=O)O"),
]

def submit_and_wait(name, smiles):
    job = client.compute(smiles=smiles, method="gfn2-xtb", tasks=["optimize"])
    result = client.wait(job.id)
    return {"name": name, "smiles": smiles, "energy": result.energy}

results = []
with ThreadPoolExecutor(max_workers=4) as executor:
    futures = {executor.submit(submit_and_wait, n, s): n for n, s in smiles_list}
    for future in as_completed(futures):
        results.append(future.result())

df = pd.DataFrame(results)
df.to_csv("batch_energies.csv", index=False)
print(df)

Key Parameters

| Parameter | Module | Default | Range / Options | Effect | |-----------|--------|---------|-----------------|--------| | method | compute | required | "gfn2-xtb", "b3lyp", "wb97x-d", "mp2" | QC method; xTB is fastest, DFT is accurate, MP2 for correlation | | basis_set | compute (DF