Correction to: Scientific Reports 10.1038/s41598-019-47148-x, published online 24 July 2019
This Article contains errors.
Penalized logP is reported in this Article for our methods in a non-normalised form. Previous research (which we compared to in the original Tables 1 and 2), starting with JT-VAE1 normalized the logP to be zero mean and unit standard deviation based on the training set. To correctly compare with previous research, our penalized logP was recalculated into a normalised form. This resulted in changes of numerical values for Penalized logP in Table 1, and numerical values for MolDQN-naïve and MolDQN-bootstrap Improvement in Table 2. The corrected versions of these tables are included below as Tables 1 and 2.
Table 1.
Top three unique molecule property scores found by each method.
| Penalized logP | QED | |||||||
|---|---|---|---|---|---|---|---|---|
| 1st | 2nd | 3rd | Validity | 1st | 2nd | 3rd | Validity | |
| random walka | −0.65 | −1.72 | −1.88 | 100% | 0.64 | 0.56 | 0.56 | 100% |
| greedyb | 9.05 | — | — | 100% | 0.39 | — | — | 100% |
| ε-greedy, ε = 0.1b | 9.10 | 9.05 | 9.05 | 100% | 0.914 | 0.910 | 0.906 | 100% |
| JT-VAEc | 5.30 | 4.93 | 4.49 | 100% | 0.925 | 0.911 | 0.910 | 100% |
| ORGANc | 3.63 | 3.49 | 3.44 | 0.4% | 0.896 | 0.824 | 0.820 | 2.2% |
| GCPNc | 7.98 | 7.85 | 7.80 | 100% | 0.948 | 0.947 | 0.946 | 100% |
| MolDQN-naive | 8.69 | 8.68 | 8.67 | 100% | 0.934 | 0.931 | 0.930 | 100% |
| MolDQN-bootstrap | 9.01 | 9.01 | 8.99 | 100% | 0.948 | 0.944 | 0.943 | 100% |
| MolDQN-bootstrap | — | — | — | — | 0.948 | 0.948 | 0.948 | 100% |
a“random walk” is a baseline that chooses a random action for each step.
b“greedy” is a baseline that chooses the action that leads to the molecule with the highest reward for each step. “ε-greedy” follows the “random” policy with probability ε, and “greedy” policy with probability 1–ε. In contrast, the ε-greedy MolDQN models choose actions based on predicted Q-values rather than rewards.
cvalues are reported in You et al.2.
Table 2.
Mean and standard deviation of penalized logP improvement in constrained optimization tasks.
| δ | JT-VAEa | GCPNa | MolDQN-naive | MolDQN-bootstrap | ||||
|---|---|---|---|---|---|---|---|---|
| Improvement | Success | Improvement | Success | Improvement | Success | Improvement | Success | |
| 0.0 | 1.91 ± 2.04 | 97.5% | 4.20 ± 1.28 | 100% | 4.83 ± 1.30 | 100% | 4.88 ± 1.30 | 100% |
| 0.2 | 1.68 ± 1.85 | 97.1% | 4.12 ± 1.19 | 100% | 3.79 ± 1.32 | 100% | 3.80 ± 1.30 | 100% |
| 0.4 | 0.84 ± 1.45 | 83.6% | 2.49 ± 1.30 | 100% | 2.34 ± 1.18 | 100% | 2.44 ± 1.25 | 100% |
| 0.6 | 0.21 ± 0.71 | 46.4% | 0.79 ± 0.63 | 100% | 1.40 ± 0.92 | 100% | 1.30 ± 0.98 | 100% |
δ is the threshold of the similarity constraint SIM(m,m0) ≥ δ. The success rate is the percentage of molecules satisfying the similarity constraint.
avalues are reported in You et al.2.
Additionally, in the Discussion, subsection “Constrained optimization”:
“Using Welch’s t-test30 for N = 800 molecules, we found that both variants of MolDQN gives a highly statistically significant improvement over GCPN for all values of δ with t < −8. The bootstrap variant also significantly outperforms the naive model (except for δ = 0.2) with t < −3.”
should read:
“Using Welch’s t-test30, we found that on δ = 0.2, both variants of MolDQN gives a statistically significant lower improvement comparing to GCPN with P < 1e-7; on δ = 0.4, the differences are insignificant at a 1% level with P = 0.016 for MolDQN naive and P = 0.43 for MolDQN bootstrap; on δ = 0.0 and 0.6, both variants of MolDQN gives a statistically significant higher improvement comparing to GCPN with P < 1e-22. The differences between two variants of MolDQN are statistically insignificant at a 1% level with P > 0.036.”
Furthermore, Figure S7 was updated. The corrected version is shown below as Figure 1.
Figure 1.
.
Lastly, References 13 and 18 were incorrectly given as:
13. Jin, W., Barzilay, R. & Jaakkola, T. Junction tree variational autoencoder for molecular graph generation. arXiv preprint arXiv:1802.04364 (2018).
18. You, J., Liu, B., Ying, R., Pande, V. & Leskovec, J. Graph convolutional policy network for goal-directed molecular graph generation. arXiv preprint arXiv:1806.02473 (2018).
The correct references are listed below as references 1 and 2.
These changes do not affect the conclusions of the Article.
References
- 1.Jin, W., Barzilay, R. & Jaakkola, T. Junction Tree Variational Autoencoder for Molecular Graph Generation. In Proceedings of the 35th International Conference on Machine Learning (eds. Dy, J. & Krause, A.) vol. 80, 2323–2332 (PMLR, 2018).
- 2.You, J., Liu, B., Ying, Z., Pande, V. & Leskovec, J. Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation. In Advances in Neural Information Processing Systems 31 (eds. Bengio, S. et al.) 6410–6421 (Curran Associates, Inc., 2018).