Skip to content

Quick Start

The mlconfgen npm package runs the ONNX weights with ONNX Runtime — no Python or PyTorch required. It mirrors the main generation path of MLConformerGeneratorONNX: context → EDM samples → AdjMatSeer bonds → standardisation and validity filtering.

Install

npm install mlconfgen onnxruntime-node

The package bundles @rdkit/rdkit (used for canonical atom ordering and sanitisation) and expects you to supply the ONNX Runtime. Node 18+.

Download the ONNX weights from Hugging Face and keep them next to your application (see Model Weights):

  • egnn_chembl_15_39.onnx
  • adj_mat_seer_chembl_15_39.onnx

Generate

import { createGenerator, seed } from "mlconfgen";
import * as ort from "onnxruntime-node";

seed(42);

const gen = await createGenerator({
  ort,
  egnnOnnx: "./egnn_chembl_15_39.onnx",
  adjMatSeerOnnx: "./adj_mat_seer_chembl_15_39.onnx",
  diffusionSteps: 100,
});

const mols = await gen.generateConformers({
  referenceContext: [89.87, 210.78, 217.78], // principal MOI components
  nAtoms: 20,
  nSamples: 10,
  variance: 2,
});

for (const mol of mols) {
  console.log(`atoms=${mol.nAtoms} bonds=${mol.bonds.length}`);
  console.log(mol.toMolBlock());
}

createGenerator(options) loads both ONNX sessions and returns an MLConformerGenerator. generateConformers returns an array of Molecule objects that passed validity filtering.

Reference from Coordinates

Instead of supplying the MOI context, pass heavy-atom coordinates and let the package compute it:

const mols = await gen.generateConformers({
  referenceConformer: { positions: flatXyzFloat32 }, // Float32Array of length n*3
  nSamples: 10,
});

positions may be a flat array, a Float32Array, or an array of [x, y, z] triples. nAtoms defaults to the number of points.

Reproducibility

seed(n) seeds a NumPy-compatible RandomState (MT19937) used for all noise draws. With the same seed and weights, randn draws match np.random.seed(n) in Python, so runs are comparable across the two implementations. Unseeded runs are initialised from crypto entropy.

Run the Examples

From a repository checkout:

cd js
npm install
npm run example                 # examples/generate.js
STEPS=50 node examples/animate.js   # writes trajectory.xyz

Both examples read EGNN_ONNX / ADJ_ONNX environment variables for weight paths.

Tests and smoke UI

npm test          # fast unit tests
npm run test:slow # ONNX generation (needs weights)
npm run test:all

npm run smoke     # local trajectory viewer at http://localhost:3847

Next