VQATOM ENCODING {html.escape(smiles)}
{draw_svg(smiles, tokens)}
PER-ATOM TOKEN SEQUENCE {chips}
Turn molecular environments into discrete tokens. Enter a SMILES string to assign a learned, context-dependent VQAtom ID to each heavy atom.
Try VQAtom Use the default example or replace it with any valid SMILES string.
Acetanilide · CC(=O)NC1=CC=CC=C1
''')
with gr.Row(elem_classes="input-card"):
smiles = gr.Textbox(value="CC(=O)NC1=CC=CC=C1", label="SMILES", lines=1, scale=5)
button = gr.Button("Generate IDs", elem_classes="encode-btn", scale=1)
output = gr.HTML(elem_classes="output-html")
gr.HTML('''
How the IDs are created
01 · GRAPH CONTEXT Aggregate the neighborhood A graph neural network collects the structural environment surrounding each atom over three hops.
02 · VECTOR QUANTIZATION Find the nearest code The continuous representation is matched to the nearest entry in an element-partitioned codebook.
VQ 1842 03 · DISCRETE TOKEN Return a global VQAtom ID The selected codebook entry becomes a reusable discrete token for that atom.
Method and evaluation: VQ-Atom: Semantic Discretization of Local Atomic Environments for Molecular Representation Learning ↗
''')
button.click(encode_molecule, smiles, output)
smiles.submit(encode_molecule, smiles, output)
if __name__ == "__main__":
demo.launch(ssr_mode=False)