Guidance

Sep 07, 2026

Feynman-Kac steering

Sep 07, 2026

Hard constraints can be enforced during diffusion by applying to fully denoised predictions and renoising

Aug 25, 2026

Annealing epitope guidance reduces steric clashes in de novo antibody design

Jul 28, 2026

Protein backbone design diffusion models can be repurposed for fitting structures into electron density

Jul 22, 2026

Distograms can be used as loss functions for protein design by hallucination

Jul 22, 2026

Effective sample size

Jul 20, 2026

Stronger diffusion guidance reduces diversity of generated outputs

Jul 19, 2026

Inversion of protein folding neural networks

Jul 19, 2026

Straight-through estimator

Jul 17, 2026

Selectivity, specificity, and conformational constraints can be encoded in loss functions during protein design

Apr 21, 2026

Diffusion-based structure prediction can be steered by modifying the conditioning embeddings rather than the latent space, and such embeddings can be used for subsequent iterations

Apr 21, 2026

Alternate conformations can be sampled by inverting AlphaFold2 and backpropagating along specific user-picked collective variables

Apr 21, 2026

Diffusion outperforms hallucination when using AF3-generation protein structure prediction methods

Apr 21, 2026

Enhanced diffusion with metadynamics-like potentials can sometimes convergence slower than unbiased diffusion

Apr 21, 2026

Ensembles can be modeled by structure prediction NNs using experimental data via guidance sampling

Apr 21, 2026

Guided sequence-only diffusion outperforms autoregressive fine-tuned PLMs on designing high-fitness sequences

Apr 21, 2026

Multimodal sequence-structure guidance outperforms unimodal guidance in cases where both need to be designed

Apr 21, 2026

Multistate Bennett acceptance ratio can be used to reweigh one or more samples from a guided diffusion trajectory

Apr 21, 2026

Ordered SAEs are more steerable than top-k SAEs

Apr 21, 2026

Protein language models can be steered to design proteins with specific properties

Apr 21, 2026

Confidence metrics for diffusion-based structure prediction methods can be improved with minimal changes to conditioning representations

Apr 21, 2026

Guidance potentials can be added to diffusion-based structure prediction for enhanced sampling of protein conformations

Apr 21, 2026

Hallucination of CDRs

Apr 21, 2026

Paratope losses are required to enforce CDR-mediated antigen binding during de novo antibody design by hallucination

Apr 21, 2026

Secondary structure losses are required to enforce CDR loopiness during de novo antibody design by hallucination

Apr 20, 2026

Diffusion guidance

inference/guidance

Quartz 4

Explorer

Guidance

Feynman-Kac steering

Hard constraints can be enforced during diffusion by applying to fully denoised predictions and renoising

Annealing epitope guidance reduces steric clashes in de novo antibody design

Protein backbone design diffusion models can be repurposed for fitting structures into electron density

Distograms can be used as loss functions for protein design by hallucination

Effective sample size

Stronger diffusion guidance reduces diversity of generated outputs

Inversion of protein folding neural networks

Straight-through estimator

Selectivity, specificity, and conformational constraints can be encoded in loss functions during protein design

Diffusion-based structure prediction can be steered by modifying the conditioning embeddings rather than the latent space, and such embeddings can be used for subsequent iterations

Alternate conformations can be sampled by inverting AlphaFold2 and backpropagating along specific user-picked collective variables

Diffusion outperforms hallucination when using AF3-generation protein structure prediction methods

Enhanced diffusion with metadynamics-like potentials can sometimes convergence slower than unbiased diffusion

Ensembles can be modeled by structure prediction NNs using experimental data via guidance sampling

Guided sequence-only diffusion outperforms autoregressive fine-tuned PLMs on designing high-fitness sequences

Multimodal sequence-structure guidance outperforms unimodal guidance in cases where both need to be designed

Multistate Bennett acceptance ratio can be used to reweigh one or more samples from a guided diffusion trajectory

Ordered SAEs are more steerable than top-k SAEs

Protein language models can be steered to design proteins with specific properties

Confidence metrics for diffusion-based structure prediction methods can be improved with minimal changes to conditioning representations

Guidance potentials can be added to diffusion-based structure prediction for enhanced sampling of protein conformations

Hallucination of CDRs

Paratope losses are required to enforce CDR-mediated antigen binding during de novo antibody design by hallucination

Secondary structure losses are required to enforce CDR loopiness during de novo antibody design by hallucination

Diffusion guidance