Pith. sign in

REVIEW 4 major objections 8 minor 63 references

Learning Disentangled Equivariant Representation for Explicitly Controllable 3D Molecule Generation

T0 review · 4 major / 8 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An equivariant Wasserstein autoencoder splits molecule latents into property and structure channels, making 3D generation explicitly controllable.

desk verdict Context-preserving generation rests on an undefined latent-combination operation and a partly circular metric; the property-targeting results are plausible but statistically weak—needs major revision before its central claim is verified. read the letter →

arxiv 2412.15086 v1 pith:UWK6LIET submitted 2024-12-19 cs.LG cs.AI

Haoran Liu , Youzhi Luo , Tianxiao Li , James Caverlee , Martin Renqiang Min This is my paper · ORCID
classification cs.LGcs.AI
keywords 3DmoleculegenerationdisentangledrepresentationlearningequivariantneuralnetworksWassersteinautoencodercontrollabledrugdesignfragment-based
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper aims to make 3D drug-like molecule generation explicitly controllable: a user should be able to demand a target property value, or demand that the overall scaffold of a molecule survive while a property changes. It proposes E3WAE, an E(3)-equivariant Wasserstein autoencoder whose latent space is factorized into a property variable $z_p$ and a structural-context variable $z_s$. A Wasserstein/MMD regularization enforces independence between the two channels, and a prediction head makes $z_p$ carry the requested property, so the decoder can generate from either channel alone. The paper reports that this yields the best property-targeting error on 7 of 8 de-novo settings and enables context-preserving generation that retrieval and diffusion baselines cannot match, including structure-based drug design against protein pockets. A correct version of this claim would give drug designers a generative model with two separate dials: one for the property, one for the shape.

What carries the argument

The central object is the factorized latent code $(z_p, z_s)$ inside a Wasserstein autoencoder: $z_p$ is pinned to a molecular property by a supervised prediction head, while $z_s$ is pushed, together with $z_p$, toward an isotropic Gaussian prior by an MMD-based Wasserstein loss, which forces the two channels to be independent. Two E(3)-equivariant encoders (built from Vector Neurons and Mixed-Features Message Passing) produce the latents, and an autoregressive fragment decoder reconstructs the molecule. The novel coordinate loss aligns generated coordinates to ground truth with an optimal rigid alignment when fewer than four nodes are present, which is what lets an equivariant decoder generate molecules de novo without external reference coordinates.

What would settle it

Take a molecule pair with the same scaffold but very different target properties (e.g., logP); swap only the property latents and generate. If the measured property of the generated molecules does not track the injected $z_p$, or if the scaffold similarity to the original falls below retrieval baselines when measured with an external fingerprint, the disentanglement claim is falsified.

Watch

Extended reading notes

Core claim

E3WAE encodes a fragment-based 3D molecular graph with two E(3)-equivariant GNN encoders into disentangled latents: the property latent $z_p$ (invariant and equivariant parts) supervised by an auxiliary property-prediction head, and the structural-context latent $z_s$ regularized by the Wasserstein autoencoder objective. The total loss $L_{\mathrm{Total}} = L_{\mathrm{Prop}} + \alpha L_{\mathrm{Dis}} + \beta L_{\mathrm{Recon}}$ aligns the joint latent distribution with an isotropic Gaussian via MMD, which makes $z_p$ and $z_s$ independent; the decoder reconstructs the molecule fragment by fragment in an autoregressive focus-and-expand procedure. To make equivariant networks work from scratch without a reference structure, the coordinate loss aligns predicted and ground-truth coordinates with an optimal rigid alignment when the generated subgraph has three or fewer nodes, preserving E(3)-equivariance of coordinates and invariance of likelihood. With this machinery the paper demonstrates property-targeting generation on GEOM-Drugs and CrossDocked2020 (best MSE/MAE on 7 of 8 settings), context-preserving generation via latent exchange, multi-property targeting, and property-guided ligand generation against protein targets with AutoDock Vina scores.

Load-bearing premise

The claim rests on the assumption that the context latent $z_s$ actually captures the molecule's structure independently of the target property, so that combining $z_s$ from one molecule with $z_p$ from another yields a valid molecule whose scaffold survives the swap.

Editorial extensions

If this is right

  • Setting only $z_p$ and sampling $z_s$ should produce molecules whose measured properties track the requested value, as reported by lower MSE/MAE than diffusion baselines on 7 of 8 property-targeting settings.
  • Setting only $z_s$ and varying $z_p$ should keep the molecular scaffold while shifting properties, which is the new context-preserving generation mode the paper demonstrates through embedding and fingerprint similarities.
  • The same factorization applies to structure-based drug design: a pocket-conditioned model can target improved drug-like properties while preserving binding, consistent with the reported Vina score improvements.
  • Because the latents are regularized to a Gaussian, interpolation and extrapolation in $z_p$ or $z_s$ become meaningful operations for steering generation.
  • The alignment coordinate loss gives a recipe for training E(3)-equivariant autoregressive decoders without reference coordinates, which should transfer to other de-novo geometric generation tasks.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the disentanglement holds beyond the paper's evaluation, the context channel $z_s$ could serve as a scaffold embedding for similarity search, fragment hopping, and lead optimization, since it is trained to be property-agnostic.
  • A natural stress test the paper does not run: generate from $z_s$ of molecule A and $z_p$ of molecule B, then measure the target property with an independent predictor (not the training head) and scaffold similarity with an external fingerprint; this would separate true disentanglement from a similarity computed in the model's own latent space.
  • The framework could extend to jointly controlling more than two factors, such as splitting $z_p$ into per-property sub-channels or adding protein-structure context, since the Wasserstein regularization is factor-agnostic.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 8 minor

Summary. The paper proposes E3WAE, an E(3)-equivariant Wasserstein autoencoder for 3D molecule generation with a latent space factorized into property variables (zp) and structural-context variables (zs). Two generation modes are claimed: property-targeting, where a desired property latent is combined with a context latent to generate a molecule with that property, and context-preserving, where a template's context latent is combined with another molecule's property latent to generate a molecule preserving the template's architecture. The method uses fragment-based auto-regressive decoding with a novel alignment-based coordinate loss. Experiments on GEOM-Drugs and CrossDocked2020 compare property-targeting accuracy against EDM, HierDiff, and TargetDiff, and context-preserving similarity against retrieval baselines.

Significance. If the central claims are supported, the paper would offer a notable step toward explicit control in 3D molecule generation, especially the context-preserving mode, which is genuinely novel relative to existing diffusion- and autoregressive-based approaches. The alignment-based coordinate loss is a useful technical contribution for adapting equivariant networks to de-novo generation, and the property-targeting point estimates outperform baselines on most settings in Tables 1-3. However, the context-preserving claim, which is the paper's unique selling point, is not fully substantiated: the cross-molecule latent combination is undefined, and the primary similarity metric is partly circular. The theoretical disentanglement guarantee in Appendix 8 also has gaps. These issues are load-bearing and need to be addressed before the claims can be accepted.

major comments (4)
  1. [Section 4.5] The cross-molecule latent combination is undefined. The encoders in Eq. (1) produce per-node latent vectors, and the decoder in Section 4.3 consumes per-node latents (e.g., in NodeTypes and Eq. (8)). For context-preserving generation, the paper says to combine the property latent of one molecule with the context latent of another (Sections 4.5 and 5.1). Since the two molecules generally have different numbers of fragments, the per-node vectors zh,p and zh,s cannot be concatenated elementwise without a matching, padding, or aggregation procedure. No such procedure is described, so the context-preserving generation mode cannot be executed as written. The authors must specify how the combination is performed and how the decoder handles latent vectors of mismatched lengths.
  2. [Section 5.2, Table 4] The evaluation of context preservation is partly circular. The primary metric is cosine similarity between the template's structural latent and the generated molecule's structural latent; because the decoder is conditioned on the template's zs, this metric is expected to be high by construction and does not independently confirm structural preservation. The fingerprint similarity is an independent signal, but in the GEOM dataset the reported Ours Fp. values (0.255, 0.218, 0.238, 0.194) are all below the retrieval maximum (0.317, 0.320, 0.318, 0.311), and the text's claim that generated molecules 'even exceed the maximum retrieval values' is supported only for the embedding metric in one setting. Please add external structural metrics (e.g., scaffold Tanimoto, graph edit distance, RMSD of fragment coordinates) and clarify the retrieval baseline procedure.
  3. [Appendix 8] The disentanglement guarantee is not rigorous. Equation (20) is used to claim that minimizing DKL(Qz||Pz) maximizes I(z;x), but the right-hand side also contains E[DKL(Qθ(z|x)||Pz)], which is not constant under the training objective; the conclusion does not follow. In addition, the MMD loss enforces marginal matching only, so the statement that I(zs;x) is 'inherently maximized by the objective' is unsupported. Please either provide a correct derivation or soften the claim to 'encourages' rather than 'guarantees.'
  4. [Tables 1-3, Appendix 11.3] The property-targeting claim of 'best performance on 7 of 8 settings' is not supported by significance testing. In Table 6, the standard deviations overlap heavily between the proposed method and baselines (e.g., GEOM Asphericity MSE: Ours 0.095±0.075 vs. HierDiff 0.176±0.096; SAS MSE: Ours 1.563±2.215 vs. HierDiff 2.618±1.908). No paired tests, confidence intervals, or effect sizes are reported. Please add statistical tests or at least discuss the overlap.
minor comments (8)
  1. [Section 1] There is a typo in 'strutcure-based' in the contribution list.
  2. [Section 4.5] In the description of context-preserving generation, 'zh,s, zv,s' for the property latent should likely be 'zh,p, zv,p' to match the notation used elsewhere.
  3. [Section 4.3, Eq. (6)] The sum over j∈Vt of ẑ_j^h appears twice in the argument of Φ; one of these terms is likely intended to aggregate equivariant features or something different.
  4. [Appendix 11.3] Table 6 is not referenced in the main text when discussing Table 1; please add a cross-reference so readers know standard deviations are available.
  5. [Figure 2] The claim that property labels are not represented by context embeddings is based on visual inspection of t-SNE plots; a quantitative measure (e.g., mutual information or R² of a linear probe) would be more convincing.
  6. [Appendix 11.4, Table 7] On GEOM-Drugs, Ours has a much larger ∆MW (69.085) than HierDiff (19.781); the statement that the model shows 'close proximity to SOTA methods in others' is misleading without a discussion of this deviation.
  7. [Section 5.2] The phrase 'we fuse the context latent embeddings' is vague; please define the fusion operation explicitly.
  8. [Throughout] The term 'fragmentization' is nonstandard; 'fragmentation' is more common in the chemistry literature.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: property targets and structural fingerprints are measured externally; the context-latent similarity metric is complemented by fingerprint similarity, and the self-cited disentanglement identity is standard.

full rationale

The paper's central claims are empirically grounded in external measurements, not in its own fitted latents. Property-targeting generation (Section 5.1) is evaluated by externally computed property values on generated molecules (e.g., QED, SA, logP) against reference values, and the property latent is trained with a supervised prediction head; the reported MSE/MAE therefore test whether the decoder can convert zp into externally verifiable properties, not whether a fitted parameter is being renamed as a prediction. Context-preserving generation (Section 5.2) does use cosine similarity in the model's own structural latent space as one metric, which is a same-space consistency check; however, the same table also reports fingerprint similarity (Fp.), an external structural descriptor, and the retrieval baselines are encoded with the same Θs, so the comparison is a baseline rather than a construction-level tautology. The re-encoded zs of a decoded molecule is not guaranteed to equal the conditioning zs; the autoencoder can fail on cross-molecule combinations, so the metric is falsifiable. The Appendix 8 'disentanglement guarantee' invokes theorems from the authors' prior work (Han et al. 2021; Li et al. 2023), but the invoked identity — DKL(Qz||Pz) = EPD[DKL(Qθ(z|x)||Pz)] - I(z;x) — is a standard information-theoretic identity that can be verified independently and does not smuggle in the paper's conclusion. The undefined cross-molecule latent combination and notation slips in Section 4.5 are reproducibility/correctness limitations, not circular reasoning. Therefore no step in the derivation reduces by construction to its input.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The model introduces no new physical entities. It introduces latent variables zp and zs, which are learned constructs supervised by property labels and reconstruction, not independently evidenced postulates. The central methodological debts are to prior E(3)-equivariant architectures, Wasserstein autoencoders, and fragment-based decoding.

free parameters (4)
  • Wasserstein loss weight α = searched over {3,5,10} on GEOM, {5,10,15} on CrossDocked
    Controls the strength of MMD disentanglement; selected by validation performance.
  • Reconstruction loss weight β = coordinate loss weight searched over {0.5,1,3}
    Balances reconstruction against disentanglement and property losses.
  • Property prediction loss weight = searched over {3,5,10,15}
    Listed in Table 5 but absent from Eq. (9), which writes LTotal = LProp + αLDis + βLRecon; this inconsistency obscures the actual objective.
  • MMD kernel bandwidth σ = 1
    RBF kernel bandwidth set to 1 by hand; no sensitivity analysis is reported.
assumptions (5)
  • domain assumption Fragment-based tree decomposition (JT-VAE) with fragment centroids as 3D coordinates adequately represents drug-like molecules
    Adopted from HierDiff/JT-VAE in Appendix 7; if fragmentization loses geometric or chemical information, generation quality is capped by this representation.
  • domain assumption The focus-expand autoregressive decoder with BFS teacher forcing can reconstruct every fragment graph in the vocabulary
    Assumed in Section 4.3; no proof of coverage, ordering invariance, or validity of the generated graphs.
  • standard math MMD minimization with RBF kernel drives the aggregate posterior to an isotropic Gaussian, enforcing independence between latent dimensions
    Only true in the infinite-data limit for a characteristic kernel; with finite batches and a fixed bandwidth it is approximate, so the disentanglement guarantee in Appendix 8 is heuristic.
  • standard math The variational identity DKL(Qz||Pz) = E[DKL(Q(z|x)||Pz)] - I(z;x) justifies maximizing I(z;x)
    The identity is standard, but the paper minimizes MMD rather than KL, so the application to the actual training objective is not direct.
  • standard math Kabsch alignment gives a unique reference frame for the coordinate loss when the current subgraph has ≤3 nodes
    Two-node subgraphs determine only a distance and three-node subgraphs only a distance and angle; Kabsch aligns predicted to ground truth. For larger subgraphs the paper uses log-MSE directly, relying on equivariance of the network outputs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Disentangled Equivariant Representation for Explicitly Controllable 3D Molecule Generation." pith.science (2026). https://pith.science/paper/UWK6LIET

@misc{pith2026241215086,
  author       = {Pith},
  title        = {Pith review of: Learning Disentangled Equivariant Representation for Explicitly Controllable 3D Molecule Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UWK6LIET}},
  note         = {Machine review of arXiv:2412.15086}
}
read the original abstract

We consider the conditional generation of 3D drug-like molecules with \textit{explicit control} over molecular properties such as drug-like properties (e.g., Quantitative Estimate of Druglikeness or Synthetic Accessibility score) and effectively binding to specific protein sites. To tackle this problem, we propose an E(3)-equivariant Wasserstein autoencoder and factorize the latent space of our generative model into two disentangled aspects: molecular properties and the remaining structural context of 3D molecules. Our model ensures explicit control over these molecular attributes while maintaining equivariance of coordinate representation and invariance of data likelihood. Furthermore, we introduce a novel alignment-based coordinate loss to adapt equivariant networks for auto-regressive de-novo 3D molecule generation from scratch. Extensive experiments validate our model's effectiveness on property-guided and context-guided molecule generation, both for de-novo 3D molecule design and structure-based drug discovery against protein targets.

Figures

Figures reproduced from arXiv: 2412.15086 by the authors.

Figure 1
Figure 1. An illustration of the proposed E3WAE framework. A 3D molecule is encoded into two disentangled latent variables: [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. t-SNE visualization of the model’s disentangled [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. An illustration of the reconstruction process. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparison of training coordinate loss curves: proposed coordinate loss vs. original log-MSE loss without structural [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Visualized 3D conformations generated by EDM. [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]
Figure 6
Figure 6. Figure 6: Visualized 3D conformations generated by HierDiff. [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Visualized 3D conformations generated by E3WAE. [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 47 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Adams, K.; and Coley, C. W. 2023. Equivariant Shape-Conditioned Generation of 3D Molecules for Ligand-Based Drug Design. In The Eleventh ICLR

    work page 2023

  4. [4]

    Axelrod, S.; and Gomez-Bombarelli, R. 2022. GEOM, energy-annotated molecular conformations for property prediction and molecular generation. Scientific Data, 9(1): 185

    2022

  5. [5]

    Barber, D.; and Agakov, F. 2004. The im algorithm: a variational approach to information maximization. NeurIPS, 16(320): 201

    work page 2004

  6. [6]

    P.; Simm, G.; Ortner, C.; and Cs \'a nyi, G

    Batatia, I.; Kovacs, D. P.; Simm, G.; Ortner, C.; and Cs \'a nyi, G. 2022. MACE: Higher order equivariant message passing neural networks for fast and accurate force fields. NeurIPS, 35: 11423--11436

    work page 2022

  7. [7]

    Bengio, E.; Jain, M.; Korablyov, M.; Precup, D.; and Bengio, Y. 2021. Flow network based generative models for non-iterative diverse candidate generation. NeurIPS, 34: 27381--27394

    work page 2021

  8. [8]

    Bengio, Y.; Courville, A.; and Vincent, P. 2013. Representation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence, 35(8): 1798--1828

    2013

Show all 63 references
  1. [9]

    H.; and Vaucher, A

    Brown, N.; Fiscato, M.; Segler, M. H.; and Vaucher, A. C. 2019. GuacaMol: benchmarking models for de novo molecular design. Journal of chemical information and modeling, 59(3): 1096--1108

    2019

  2. [10]

    P.; Higgins, I.; Pal, A.; Matthey, L.; Watters, N.; Desjardins, G.; and Lerchner, A

    Burgess, C. P.; Higgins, I.; Pal, A.; Matthey, L.; Watters, N.; Desjardins, G.; and Lerchner, A. 2018. Understanding disentangling in -VAE. arXiv preprint arXiv:1804.03599

    2018 arXiv

  3. [11]

    Ceylan, C.; and Gutmann, M. U. 2018. Conditional noise-contrastive estimation of unnormalised models. In ICML, 726--734. PMLR

    2018

  4. [12]

    Chen, X.; Duan, Y.; Houthooft, R.; Schulman, J.; Sutskever, I.; and Abbeel, P. 2016. Infogan: Interpretable representation learning by information maximizing generative adversarial nets. NeurIPS, 29

    2016

  5. [13]

    R.; Shen, D.; Malon, C.; Zhang, Y.; Li, Y.; and Carin, L

    Cheng, P.; Min, M. R.; Shen, D.; Malon, C.; Zhang, Y.; Li, Y.; and Carin, L. 2020. Improving Disentangled Text Representation Learning with Information-Theoretic Guidance. In Jurafsky, D.; Chai, J.; Schluter, N.; and Tetreault, J., eds., Proceedings of the 58th Annual Meeting ...

    2020

  6. [14]

    Cover, T. M. 1999. Elements of information theory. John Wiley & Sons

    1999

  7. [15]

    Deng, C.; Litany, O.; Duan, Y.; Poulenard, A.; Tagliasacchi, A.; and Guibas, L. J. 2021. Vector neurons: A general framework for so (3)-equivariant networks. In ICCV, 12200--12209

    2021

  8. [16]

    Deng, Y.; Bakhtin, A.; Ott, M.; Szlam, A.; and Ranzato, M. 2020. Residual Energy-Based Models for Text Generation. In ICLR

    2020

  9. [17]

    L.; et al

    Denton, E. L.; et al. 2017. Unsupervised learning of disentangled representations from video. NeurIPS, 30

    2017

  10. [18]

    Du, Y.; Guo, X.; Wang, Y.; Shehu, A.; and Zhao, L. 2022. Small molecule generation via disentangled representation learning. Bioinformatics, 38(12): 3200--3208

    2022

  11. [19]

    F.; and Forli, S

    Eberhardt, J.; Santos-Martins, D.; Tillack, A. F.; and Forli, S. 2021. AutoDock Vina 1.2. 0: New docking methods, expanded force field, and python bindings. Journal of chemical information and modeling, 61(8): 3891--3898

    2021

  12. [20]

    Fey, M.; and Lenssen, J. E. 2019. Fast Graph Representation Learning with PyTorch Geometric . In ICLR Workshop on Representation Learning on Graphs and Manifolds

    2019

  13. [21]

    G.; Masuda, T.; Sunseri, J.; Jia, A.; Iovanisci, R

    Francoeur, P. G.; Masuda, T.; Sunseri, J.; Jia, A.; Iovanisci, R. B.; Snyder, I.; and Koes, D. R. 2020. Three-dimensional convolutional neural networks and a cross-docked data set for structure-based drug design. Journal of chemical information and modeling, 60(9): 4200--4215

    2020

  14. [22]

    M.; and Sun, J

    Fu, T.; Xiao, C.; Li, X.; Glass, L. M.; and Sun, J. 2021. Mimosa: Multi-constraint molecule sampling for molecule optimization. In AAAI, volume 35, 125--133

    2021

  15. [23]

    Garcia Satorras, V.; Hoogeboom, E.; Fuchs, F.; Posner, I.; and Welling, M. 2021. E (n) equivariant normalizing flows. NeurIPS, 34: 4181--4192

    2021

  16. [24]

    Gebauer, N.; Gastegger, M.; and Sch \"u tt, K. 2019. Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules. NeurIPS, 32

    2019

  17. [25]

    Geiger, M.; and Smidt, T. 2022. e3nn: Euclidean Neural Networks

    2022

  18. [26]

    M.; Rasch, M

    Gretton, A.; Borgwardt, K. M.; Rasch, M. J.; Sch \"o lkopf, B.; and Smola, A. 2012. A kernel two-sample test. The Journal of Machine Learning Research, 13(1): 723--773

    2012

  19. [27]

    W.; Peng, X.; Su, Y.; Peng, J.; and Ma, J

    Guan, J.; Qian, W. W.; Peng, X.; Su, Y.; Peng, J.; and Ma, J. 2023. 3D Equivariant Diffusion for Target-Aware Molecule Generation and Affinity Prediction. In The Eleventh ICLR

    2023

  20. [28]

    R.; Han, L.; Li, L

    Han, J.; Min, M. R.; Han, L.; Li, L. E.; and Zhang, X. 2021. Disentangled Recurrent Wasserstein Autoencoder. In ICLR

    2021

  21. [29]

    Higgins, I.; Matthey, L.; Pal, A.; Burgess, C.; Glorot, X.; Botvinick, M.; Mohamed, S.; and Lerchner, A. 2016. beta-vae: Learning basic visual concepts with a constrained variational framework. In ICLR

    2016

  22. [30]

    G.; Vignac, C.; and Welling, M

    Hoogeboom, E.; Satorras, V. G.; Vignac, C.; and Welling, M. 2022. Equivariant diffusion for molecule generation in 3d. In ICML. PMLR

    2022

  23. [31]

    Huang, H.; Sun, L.; Du, B.; and Lv, W. 2023. Learning Joint 2D & 3D Diffusion Models for Complete Molecule Generation. arXiv preprint arXiv:2305.12347

    2023 arXiv

  24. [32]

    Huang, Y.; Peng, X.; Ma, J.; and Zhang, M. 2022. 3DLinker: an E (3) equivariant variational autoencoder for molecular linker design. arXiv preprint arXiv:2205.07309

    2022 arXiv

  25. [33]

    G.; Frossard, P.; Welling, M.; Bronstein, M.; and Correia, B

    Igashov, I.; St \"a rk, H.; Vignac, C.; Satorras, V. G.; Frossard, P.; Welling, M.; Bronstein, M.; and Correia, B. 2022. Equivariant 3d-conditional diffusion models for molecular linker design. arXiv preprint arXiv:2210.05274

    2022 arXiv

  26. [34]

    R.; van der Schaar, M.; and Deane, C

    Imrie, F.; Bradley, A. R.; van der Schaar, M.; and Deane, C. M. 2020. Deep generative models for 3D linker design. Journal of chemical information and modeling, 60(4): 1983--1995

    2020

  27. [35]

    Jin, W.; Barzilay, R.; and Jaakkola, T. 2018. Junction Tree Variational Autoencoder for Molecular Graph Generation. In ICML, 2323--2332

    2018

  28. [36]

    Kabsch, W. 1976. A solution for the best rotation to relate two sets of vectors. Acta Crystallographica Section A: Crystal Physics, Diffraction, Theoretical and General Crystallography, 32(5): 922--923

    1976

  29. [37]

    Kim, H.; and Mnih, A. 2018. Disentangling by factorising. In ICML, 2649--2658. PMLR

    2018

  30. [38]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2015. Adam: A method for stochastic optimization. In ICLR

    2015

  31. [39]

    P.; and Welling, M

    Kingma, D. P.; and Welling, M. 2014. Auto-Encoding Variational Bayes . In 2nd ICLR, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings

    2014

  32. [40]

    Kotovenko, D.; Sanakoyeu, A.; Lang, S.; and Ommer, B. 2019. Content and style disentanglement for artistic style transfer. In ICCV, 4422--4431

    2019

  33. [41]

    Landrum, G. 2016. RDKit: open-source cheminformatics http://www. rdkit. org. 3(8)

    2016

  34. [42]

    Lee, H.-Y.; Tseng, H.-Y.; Huang, J.-B.; Singh, M.; and Yang, M.-H. 2018. Diverse image-to-image translation via disentangled representations. In Proceedings of the European conference on computer vision (ECCV), 35--51

    2018

  35. [43]

    Li, T.; Guo, H.; Grazioli, F.; Gerstein, M.; and Min, M. R. 2023. Disentangled Wasserstein Autoencoder for T-Cell Receptor Engineering. In Thirty-seventh Conference on Neural Information Processing Systems

    2023

  36. [44]

    Liu, M.; Luo, Y.; Uchino, K.; Maruhashi, K.; and Ji, S. 2022. Generating 3 D Molecules for Target Protein Binding. In Chaudhuri, K.; Jegelka, S.; Song, L.; Szepesvari, C.; Niu, G.; and Sabato, S., eds., Proceedings of the 39th ICML, volume 162 of Proceedings of Machine Learnin...

    2022

  37. [45]

    Locatello, F.; Bauer, S.; Lucic, M.; Raetsch, G.; Gelly, S.; Sch \"o lkopf, B.; and Bachem, O. 2019. Challenging common assumptions in the unsupervised learning of disentangled representations. In ICML, 4114--4124. PMLR

    2019

  38. [46]

    Luo, S.; Guan, J.; Ma, J.; and Peng, J. 2021. A 3D generative model for structure-based drug design. NeurIPS

    2021

  39. [47]

    Luo, Y.; and Ji, S. 2022. An Autoregressive Flow Model for 3D Molecular Geometry Generation from Scratch. In ICLR

    2022

  40. [48]

    Mollaysa, A.; Paige, B.; and Kalousis, A. 2020. Conditional generation of molecules from disentangled representations

    2020

  41. [49]

    Paszke, A.; Gross, S.; Chintala, S.; Chanan, G.; Yang, E.; DeVito, Z.; Lin, Z.; Desmaison, A.; Antiga, L.; and Lerer, A. 2017. Automatic differentiation in pytorch

    2017

  42. [50]

    I., Madzhidov; and Varnek, A

    Polishchuk, P.; G, T. I., Madzhidov; and Varnek, A. 2013. Estimation of the size of drug-like chemical space based on GDB-17 data. Journal of computer-aided molecular design, 27: 675--679

    2013

  43. [51]

    S.; Yu, H

    Powers, A. S.; Yu, H. H.; Suriana, P. A.; and Dror, R. O. 2022. Fragment-based ligand generation guided by geometric deep learning on protein-ligand structures. In ICLR2022 Machine Learning for Drug Discovery

    2022

  44. [52]

    Qiang, B.; Song, Y.; Xu, M.; Gong, J.; Gao, B.; Zhou, H.; Ma, W.-Y.; and Lan, Y. 2023. Coarse-to-fine: a hierarchical diffusion model for molecule generation in 3d. In ICML, 28277--28299. PMLR

    2023

  45. [53]

    G.; Hoogeboom, E.; and Welling, M

    Satorras, V. G.; Hoogeboom, E.; and Welling, M. 2021. E (n) equivariant graph neural networks. In ICML, 9323--9332. PMLR

    2021

  46. [54]

    Shi, C.; Xu, M.; Zhu, Z.; Zhang, W.; Zhang, M.; and Tang, J. 2019. GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation. In ICLR

    2019

  47. [55]

    J.; Tung, H.-Y.; Strathmann, H.; De, S.; Ramdas, A.; Smola, A.; and Gretton, A

    Sutherland, D. J.; Tung, H.-Y.; Strathmann, H.; De, S.; Ramdas, A.; Smola, A.; and Gretton, A. 2017. Generative Models and Model Criticism via Optimized Maximum Mean Discrepancy. In ICLR

    2017

  48. [56]

    Tolstikhin, I.; Bousquet, O.; Gelly, S.; and Schoelkopf, B. 2018. Wasserstein Auto-Encoders. In ICLR

    2018

  49. [57]

    Van der Maaten, L.; and Hinton, G. 2008. Visualizing data using t-SNE. Journal of machine learning research, 9(11)

    2008

  50. [58]

    Xie, Y.; Shi, C.; Zhou, H.; Yang, Y.; Zhang, W.; Yu, Y.; and Li, L. 2021. \ MARS \ : Markov Molecular Sampling for Multi-objective Drug Discovery. In ICLR

    2021

  51. [59]

    S.; Dror, R

    Xu, M.; Powers, A. S.; Dror, R. O.; Ermon, S.; and Leskovec, J. 2023. Geometric latent diffusion models for 3d molecule generation. In ICML, 38592--38610. PMLR

    2023

  52. [60]

    Yang, S.; Hwang, D.; Lee, S.; Ryu, S.; and Hwang, S. J. 2021. Hit and lead discovery with explorative rl and fragment-based molecule generation. NeurIPS, 34: 7924--7936

    2021

  53. [61]

    Yu, R. 2020. A Tutorial on VAEs: From Bayes' Rule to Lossless Compression. arXiv preprint arXiv:2006.10273

    2020 arXiv

  54. [62]

    Zhu, J.-Y.; Zhang, Z.; Zhang, C.; Wu, J.; Torralba, A.; Tenenbaum, J.; and Freeman, B. 2018. Visual object networks: Image generation with disentangled 3D representations. NeurIPS, 31

    2018

  55. [63]

    R.; Kadav, A.; and Graf, H

    Zhu, Y.; Min, M. R.; Kadav, A.; and Graf, H. P. 2020. S3vae: Self-supervised sequential vae for representation disentanglement and data generation. In CVPR, 6538--6547

    2020

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.