- Rating
- 1
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
The paper proposes FLDWM, a rotation-conditioned latent denoiser for 3D molecular generation. Molecules are voxelized and encoded with a VQ-VAE. During training, the model receives a noisy latent code of a randomly rotated molecule and the applied rotation matrix, and predicts the latent code in the molecule’s original pose. At inference, generation starts from a seed molecule and uses Walk–Jump Sampling in latent space. The paper reports results on GEOM-Drugs and PCQM and claims a bound on approximate SO(3) equivariance.
Critical strengths
- Conditioning a latent denoiser on a known rotation and training it to recover the original pose is a concrete learning objective that may encourage approximate SO(3) equivariance without requiring an explicitly equivariant backbone. The formulation is interesting.
- On GEOM-Drugs at 10 steps, FLDWM reports higher validity than NEBULA at nearly identical seed similarity. The drug-seed examples further show that the pipeline can produce structurally distinct analogs across different starting molecules.
Critical weaknesses
- Presentation and positioning
- Some aspects of the paper’s framing are unclear and may mislead readers who are less familiar with this area. The paper seems to blur the distinction between seed-conditioned generation and the broader task of molecular generation. Its generation experiments require a seed molecule, but the abstract and introduction frame the contribution more broadly as 3D molecular generation. The paper should make the narrower task explicit earlier.
- The discussion of baselines also blurs this distinction. NEBULA is designed for seed-conditioned generation, whereas EDM and MiDi are not seed-conditioned in their original form. Although the authors adapt the latter methods through partial noising, these are not originally the same generation task, and the comparison should be presented with that qualification.
- The positioning relative to prior work is questionable. In terms of its overall pipeline, this work largely builds on NEBULA and adds a rotation-aware denoiser conditioned on R. However, the related-work section states: “Closest to our setting, RADM likewise pursues scalable non-equivariant generation by handling rotations explicitly, through rotational alignment.”
- Experiments
- The proposed denoiser is substantially larger than NEBULA’s. It is therefore unclear whether the improvement over NEBULA comes from scaling up the model or from the rotation-conditioned denoiser. A comparison is needed to isolate the proposed contribution.
- Relatedly, comparing the present seed-conditioned model with EDM and MiDi is difficult to interpret. The authors adapt these models through partial noising, but the noise is applied in different spaces. EDM and MiDi perturb molecular representations directly, whereas FLDWM perturbs a compressed VQ-VAE latent. Even when a latent code is substantially noised, the decoder may still produce a molecule similar to the seed. However, direct noising in molecular space may destroy seed information much more rapidly. Matching the input signal-to-noise ratio therefore does not establish comparable effective perturbations or comparable outputs. Moreover, EDM is trained to reverse a multi-step diffusion process, whereas FLDWM is trained to perform a single denoising jump after its latent-space walk. These differences make the current comparison difficult to interpret as a controlled comparison of seed-conditioned generation.
- In Table 3, the PCQM “0 (data)” row reports valency , atom TV , bond TV , bond-length , and bond-angle . The corresponding 10-step FLDWM values are , , , , and . If these distances use PCQM as their reference distribution, it is difficult to understand why the original data are so much farther from that reference than the generated molecules. One possible explanation is that the reference is instead GEOM-Drugs. If so, smaller distances would indicate greater similarity to GEOM-Drugs, rather than showing that the outputs match the PCQM test distribution. The authors need to specify the reference dataset if they want to use this table to support the cross-dataset fidelity.
- Math
- The proposed latent rotation is not shown to be a well-defined, invertible, linear group action. Because includes lossy VQ-VAE encoding, invertibility of in molecular coordinate space does not establish the corresponding properties in latent space.
- Theorem 1 assumes the decomposition , with an -independent . FiLM conditioning does not enforce this decomposition, and a low denoising loss does not establish that the FiLM layers implement the inverse latent rotation.
- Equation (5) defines an expected squared loss over molecules, rotations, and Gaussian noise. The proof then uses it as a bound for individual clean inputs in Eqs. (9) and (13). In particular, suppressing the noise term and moving from an average error to a pointwise error require justification. The resulting pointwise equivariance bound does not follow from the stated assumption.
- Equation (14) defines as a supremum over molecules and rotations, whereas Appendix A.4 reports its average over sampled pairs. That average cannot be substituted into the theorem to obtain the claimed numerical “certificate.” Moreover, the appendix acknowledges an encoder contribution that need not vanish as , which is difficult to reconcile with the stated .
Questions for authors
-
Is FLDWM intended only for seed-conditioned generation, or can it also generate molecules without a seed? If unconditional generation is within the intended scope, could the authors describe how sampling is initialized and provide corresponding results?
-
How much of the improvement remains when the only substantive difference is rotation conditioning?
-
What reference dataset is used for each distributional distance in Table 3, and how exactly is the “0 (data)” row computed?
-
What assumptions ensure that is a well-defined linear group action and that the FiLM-conditioned model has the factorization required by Theorem 1? How does an expected loss on noisy inputs yield the stated pointwise bound on clean inputs, and can the numerical certificate be recomputed using a quantity consistent with the supremum in Eq. (14)?
We thank the reviewers and the area chair for their time and careful feedback. We agree with the main concerns raised. We will substantially revise the work and resubmit.