Table 1 Performance of PGMG and other SMILES-based models

From: A pharmacophore-guided deep learning approach for bioactive molecular generation

Methods

Validity↑

Uniqueness↑

Novelty↑

Ratio of Available Molecules↑

ORGAN9

0.379

0.841

0.687

0.219

VAE4

0.870

0.999

0.974

0.847

SMILES LSTM32

0.959

1.000

0.912

0.875

Syntalinker17

1.000

0.880

0.903

0.795

PGMG

0.982

0.979

0.976

0.938

  1. An upward arrow next to each metric indicates that higher values represent better performance. Best performance among all methods for each metric is shown in bold.