- Rating
- 3
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 3
Read review
Paper summary
The paper proposes PepMALD, a structure-based peptide design framework built upon latent diffusion. The authors identify limitations in both the latent space learned by conventional VAEs and the subsequent diffusion process, and introduce three methodological improvements. First, they propose a manifold-anchored hierarchical autoencoder that aggregates multi-level representations and global context, while replacing conventional KL regularization with L2 penalties on the E(3)-invariant sequence latent representations and the deviations of E(3)-equivariant structural latents from their corresponding Cα coordinates. Second, they introduce a latent representation alignment loss that encourages intermediate denoiser representations to recover the clean sequence latents. Third, they develop an interface-guided adaptive diffusion scheme that adjusts residue-specific noise schedules according to receptor proximity, allowing interface-proximal residues to stabilize earlier during generation. The authors evaluate PepMALD on sequence-structure co-design and binding conformation generation using two benchmarks, PepBench and PepBDB. PepMALD outperforms the evaluated baselines on several metrics, with particularly notable improvements in binding energy and structural accuracy on PepBench, although the gains on PepBDB are more modest.
Critical strengths
The most notable strength of this paper lies in its comprehensive experimental evaluation and analysis. Beyond conventional benchmarks covering both de novo peptide design and binding conformation generation across two datasets, the authors construct an additional post-cutoff test set to assess generalization to newly released complexes. The paper also presents extensive analyses, including independent validation using cofolding models, molecular dynamics simulations, latent space characterization, and investigations of intermediate denoiser representations and diffusion dynamics. Together, these experiments provide a broad assessment of the proposed method's effectiveness and offer insights into how its individual components contribute to the observed improvements.
Critical weaknesses
The most significant weakness is that some methodological choices lack convincing conceptual justification. First, it is unclear why conventional KL regularization is replaced with simple L2 penalties. KL regularization encourages a continuous latent distribution compatible with a Gaussian prior, whereas L2 regularization merely constrains latent magnitudes without explicitly promoting distributional continuity or smooth interpolation. This raises concerns about the suitability of the resulting latent space for diffusion-based generation. Second, the motivation for aligning intermediate denoiser representations with clean latents is also unclear. Since the standard diffusion noise-prediction objective already implicitly supervises the recovery of clean latents through the denoiser's output, the necessity and additional benefits of imposing such supervision on intermediate hidden representations remain insufficiently justified.
Questions for authors
- What is the definition of the RMSD metric in Table 1? Is it calculated with the native peptide structure, or with the boltz/chai-predicted structure? Is it calculated on , or on all atoms?
- How does IGAD address the training–sampling discrepancy caused by using clean latents during training but noisy latents to define the distance during inference?
- How does the proposed “manifold anchoring” differ from replacing KL with conventional L2 regularization?
- How sensitive is IGAD to the interface distance threshold and the choice of adaptive noise schedule?
- Does IGAD actually cause interface residues to stabilize earlier during generation? Could the authors compare the denoising trajectories of interface residues with and without IGAD?
4 comments
4. How sensitive is IGAD to the interface distance threshold and the choice of adaptive noise schedule?
Thank you for asking about sensitivity. IGAD uses a continuous Gaussian score, , with distance and width in normalized pocket coordinates. We varied the default width 1.00 by ±25% and replaced linear with quadratic , both bounded in [0.5,1].
Using the same PepMALD weights, initial states, random perturbations and 100 sampling steps, we generated one sample per setting for each of 93 complexes.
Table 4
| Mapping | RMSD (Å) ↓ | DockQ ↑ | |
|---|---|---|---|
| 1.00 | Linear (default) | 8.555 ± 6.127 | 0.413 ± 0.230 |
| 0.75 | Linear | 8.256 ± 6.246 | 0.431 ± 0.235 |
| 1.25 | Linear | 8.206 ± 6.118 | 0.432 ± 0.232 |
| 1.00 | Quadratic | 8.687 ± 6.381 | 0.414 ± 0.236 |
Reducing/increasing width lowered RMSD by 3.5%/4.1% and increased DockQ by 0.018/0.019. Quadratic scheduling increased RMSD by 1.5% with little DockQ change. These results show that IGAD maintains similar generation performance across the tested distance parameters and schedule mappings, indicating robustness to these parameter choices.
5. Why is LRA needed in addition to noise prediction?
Thank you for requesting this distinction. Noise prediction supervises the output used to recover clean latents from noisy inputs and diffusion coefficients. LRA directly supervises hidden states to standardized sequence latents from frozen MHAE: where contains peptide residues and is the last intermediate representation. The clean reference is fixed across timesteps, encouraging consistently accessible clean-latent information.
The w/o LRA ablation (Fig. 13) retains noise prediction and evaluates this added supervision. Table 5 summarizes linear-readout results from frozen denoiser representations (Fig. 4A and Appendix H). The coefficient of determination, , measures clean-latent recovery relative to mean prediction: 1 indicates perfect recovery, 0 matches the mean baseline, and negative values indicate worse performance.
Table 5
| PepMALD | w/o LRA | |
|---|---|---|
| 90 | 0.119 | 0.140 |
| 70 | 0.297 | 0.281 |
| 50 | 0.558 | 0.551 |
| 30 | 0.851 | 0.818 |
| 10 | 0.978 | 0.951 |
PepMALD has higher recoverability at most evaluated timestep. Together with the w/o LRA ablation, this supports complementary representation supervision.
6. Does IGAD cause interface residues to stabilize earlier during generation?
Thank you for suggesting this comparison. We compared PepMALD with the w/o IGAD model on 93 test complexes using three runs per model, with matched receptor conditions, initial states, random perturbations and 100 sampling steps.
Interface residues were defined by a native minimum peptide–receptor distance below 8 Å, with grouping fixed throughout. The 92 complexes containing interface residues were included. Table 6 reports the distance to each trajectory's own final state and displacement over five steps, measured using structural latent coordinates in Å. Runs were averaged within each complex before computing the mean and standard deviation across complexes.
Table 6
| Progress | Distance, w/o IGAD | Distance, PepMALD | Displacement, w/o IGAD | Displacement, PepMALD |
|---|---|---|---|---|
| 20% | 10.022 ± 2.343 | 8.726 ± 2.134 | 7.231 ± 1.810 | 5.485 ± 1.195 |
| 40% | 7.326 ± 1.617 | 6.499 ± 1.442 | 4.594 ± 1.190 | 3.624 ± 0.853 |
| 60% | 5.235 ± 1.310 | 4.653 ± 0.978 | 3.311 ± 0.767 | 2.647 ± 0.590 |
| 80% | 2.834 ± 0.659 | 2.482 ± 0.496 | 2.329 ± 0.590 | 1.966 ± 0.420 |
| 90% | 1.542 ± 0.381 | 1.328 ± 0.304 | 1.691 ± 0.398 | 1.436 ± 0.292 |
At 20% progress, PepMALD reduced the distance to its own final state from 10.022 to 8.726 Å and five-step displacement from 7.231 to 5.485 Å relative to w/o IGAD. The same pattern held at every reported stage: interface residues were closer to their respective final states and underwent smaller positional changes, supporting earlier stabilization in these trajectories.
We will add this trajectory comparison to §4.4 after the IGAD coefficient analysis.
Thanks
Thanks for the detailed responses, which should help the readers better understand the paper. I decided to maintain the current score of weak acceptance.
We sincerely thank you for your time, thoughtful feedback, and valuable guidance throughout the review process.
1. What is the definition of the RMSD metric in Table 1?
Thank you for asking. RMSD (Å) compares generated peptide Cα coordinates with those in the native test-set receptor–peptide complex. It uses Cα atoms only and the native reference, not Boltz/Chai predictions: RMSD=N1∑i=1N∣xigen−xinative∣22 where xi denotes corresponding Cα coordinates and N counts valid residue pairs.
We will add this definition to the evaluation-metrics description accompanying Table 1 in the main text, explicitly specifying the native reference and the use of Cα atoms.
2. How does IGAD address the training–sampling discrepancy caused by using clean latents during training but noisy latents to define the distance during inference?
Thank you for highlighting this. Training defines forward noising from clean structural latents; the denoiser receives noisy latents and local time. Sampling recomputes scores from current noisy latents and the fixed receptor each reverse step. Gaussian distance scores χit continuously set Kit=1−χit(1−Kmin), with Kmin=0.5 and Kit∈[0.5,1]. The same current exponent determines local time, signal coefficients and noise variance, with scores reassessed each step.
We forward-corrupted clean latents for 93 complexes, repeated three times, and compared noisy versus clean score rankings by within-complex Spearman correlation, averaging repetitions before summarizing complexes.
Table 1
Agreement increases as corruption decreases, supporting dynamic score/schedule updates without requiring clean-reference agreement at every step.
We will clarify in §3.3 how clean anchors determine forward noising during training and how dynamically recomputed anchors jointly determine local time conditioning and reverse-update coefficients during sampling.
3. Why replace KL regularization with L2, and how does “manifold anchoring” differ from conventional L2 regularization?
Thank you for this question. MHAE learns deterministic reconstruction codes; conditional diffusion learns their distribution. A Gaussian diffusion endpoint does not require Gaussian clean codes. For q=N(μ,diag(σ2)) and p=N(m,I), DKL(q∣p)=21∣μ−m∣22+21∑j(σj2−logσj2−1) With fixed covariance, the variance term is constant, leaving a quadratic mean penalty. This motivates deterministic encoding with magnitude/offset control, leaving distribution modeling to diffusion.
VAE/MHAE checkpoints were evaluated on 93 LNR complexes; H/Z denote perturbation strengths:
Table 2
Retention uses unperturbed decoding as reference. MHAE improves clean reconstruction and sequence consistency.
Anchoring specifies geometric references: Lanch=2N1∑i(λH∣ui∣22+λZ∣zi−ai∣22) ui,zi are sequence/structural latents; ai is native Cα. H uses zero-centered L2; Z uses a residue-specific reference, invariant to joint rotation/translation and allowing learned offsets.
We trained no-Z and backbone-center controls with the same MHAE architecture and H regularization. The center averages valid N/Cα/C/O coordinates; original Cα-anchored MHAE is the reference.
Table 3
RMSD/MAE use Å; MAE uses ideal backbone bond lengths. Clashes are pairs <1.2 Å apart, excluding one/two-bond neighbors. Backbone-center versus no-Z improvements support residue-level geometric anchoring beyond H magnitude control.
We will clarify the motivation for deterministic regularization and residue-specific geometric anchoring in §3.1, and extend Appendix C.3 with the Gaussian KL decomposition and the reconstruction, perturbation, and anchoring-control results reported above.