Pith. sign in

REVIEW 4 major objections 5 minor 77 references

AffinityFlow: Guided Flows for Antibody Affinity Maturation

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read AffinityFlow reports state-of-the-art affinity maturation by alternating guided flow sampling with inverse folding and co-taught predictors.

desk verdict The pipeline is real and clearly presented, but the SOTA affinity-maturation claim is circular: Rosetta ΔG serves as both the training signal and the evaluation metric. read the letter →

arxiv 2502.10365 v2 pith:F4PA45T7 submitted 2025-02-14 cs.LG

Can Chen , Karla-Luise Herpoldt , Chenchao Zhao , Zichen Wang , Marcus Collins , Shang Shang , Ron Benson This is my paper
classification cs.LG
keywords antibodyaffinitymaturationflowmatchingpredictorguidanceco-teachinginversefoldingsequence-onlyproteindesignsingle-domainbindingprediction
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

Affinity maturation normally requires experimenters to mutate antibodies and test whether they bind their target more tightly. This paper claims that the process can be driven by sequence information alone: an antibody and an antigen sequence are enough to propose mutations that improve predicted binding. The method alternates between two steps—steering a pretrained structure-generating flow model toward low predicted binding energy, and inverse-folding those structures into mutations that a sequence-based predictor then filters. Because real affinity measurements are scarce, the paper trains both predictors with a co-teaching scheme that converts noisy Rosetta binding-energy differences on docked complexes into pairwise labels, with each predictor selecting trustworthy samples for the other. On a single-domain antibody benchmark, AffinityFlow reports the highest fraction of improved mutants and the highest antigen specificity among the compared methods.

What carries the argument

The central object is the guided flow-matching sampler. AlphaFlow is a pretrained sequence-conditioned generative model that maps noisy protein backbone coordinates to clean coordinates, and AffinityFlow steers that process with the velocity field $$\tilde{v}(x_t,t,\$\Delta$ G;\$\theta$)=\hat{v}(x_t,t;\$\theta$)-\gamma\frac{1-t}{t}\nabla_{x_t}\hat{f}_\$\beta$(\hat{x}_1(x_t)),$$ where $\hat{f}_\beta$ is the structure-based affinity predictor. The gradient term directs denoising toward lower predicted binding energy, and the guidance is applied mainly to CDR coordinates in the later sampling steps. The second load-bearing mechanism is the co-teaching module: pairwise labels are formed from Rosetta $\Delta\Delta G$ values on GeoDock-docked complexes, and the sequence-based and structure-based predictors each select consensus samples to fine-tune the other, filtering noisy labels. The loop is closed by ProteinMPNN inverse folding, Amber relaxation as a predictor-corrector step, and an ESM-2/MLP sequence-based predictor for post-selection.

What would settle it

Measure the actual binding affinities of a sample of the paper's designed mutants (for example by surface plasmon resonance or biolayer interferometry) and compare the measured ordering with the ordering predicted by Rosetta on the IgFold/GeoDock structures; if the measured mutants do not bind tighter than the wild-type or do not correlate with predicted $\Delta\Delta G$, the central claim collapses.

Watch

Extended reading notes

Core claim

The paper's central claim is that antibody affinity maturation can be formulated as an alternating optimization between structure and sequence, and that this beats mutating directly in sequence space. Starting from an antibody-antigen pair joined as one sequence, AffinityFlow freezes the sequence while a structure-based affinity predictor guides AlphaFlow's flow-matching sampler to low-energy backbone conformations; it then applies ProteinMPNN inverse folding to generate one-, two-, and three-point CDR mutations, and a sequence-based affinity predictor selects the promising variants. The two predictors are refined by a co-teaching module: Rosetta binding free energies computed on GeoDock-docked complexes are turned into pairwise discrete labels ($\Delta\Delta G$), and each predictor selects samples on which it agrees to fine-tune the other. The paper reports that this loop, run for three iterations, achieves the best Improvement Percentage (IMP) and the lowest cross-antigen sequence similarity (Sim) in Table 1 across CDR-H1, CDR-H2, CDR-H3, and all CDRs, and, among non-language-model methods, the best naturalness score.

Load-bearing premise

The load-bearing premise is that Rosetta binding free energy computed on GeoDock-docked antibody-antigen complexes is a valid proxy for true binding affinity; if these energies do not track real binding, both the guidance signal and the benchmark are measuring the wrong quantity.

Editorial extensions

If this is right

  • Antibody maturation no longer needs an experimentally solved antibody-antigen complex structure: a pair of sequences is sufficient to enter the design loop.
  • The alternating recipe can be lifted onto any sequence-conditioned structure generator: guide the structure with a differentiable affinity surrogate, inverse-fold, and filter with a sequence-ranked surrogate.
  • The co-teaching module gives a template for training affinity predictors from noisy physics-based labels when supervised affinity data are scarce.
  • Because the method generates antigen-specific mutants (low Sim), it is positioned for target-specific therapeutic design rather than generic antibody mutation.
  • The case-study mutations such as Ala105Pro and Lys99Trp against the SARS-CoV-2 RBD are concrete, testable predictions.

Reading between the lines

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

  • The reported benchmark is self-consistent: the same Rosetta energy oracle that creates the co-teaching labels also scores the final designs, so the numbers demonstrate optimization within a physics-based surrogate rather than validated wet-lab improvement.
  • The framework can be read as a model-based optimizer with two coupled surrogates—a differentiable structure predictor for coordinates and a ranking sequence predictor for mutations—so the co-teaching scheme may transfer to other design problems with noisy biophysical labels.
  • A natural next experiment is to run the same pipeline on antibodies with published experimental affinities and check whether the top Rosetta-ranked variants are also the top binders in the lab; that would separate design quality from surrogate bias.
  • The use of pairwise relative labels ($\Delta\Delta G$) rather than absolute energies matches the intuition that ranking is easier than scoring, a choice likely worth reusing even when some experimental labels exist.
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 / 5 minor

Summary. The paper presents AffinityFlow, a sequence-only antibody affinity maturation method that combines AlphaFlow flow matching with a structure-based affinity predictor for guided structure generation, followed by ProteinMPNN inverse folding and a sequence-based predictor for post-selection. The predictors are trained with a co-teaching scheme over pairwise Rosetta ΔΔG labels computed from GeoDock-docked sdAb-antigen complexes. Experiments on a SAbDab sdAb benchmark report IMP, Sim, and Nat metrics across CDR-H1, CDR-H2, CDR-H3, and all CDR positions, claiming state-of-the-art performance against ESM, AbLang, nanoBERT, dWJS, DiffAb, AbDPO, and GearBind.

Significance. The proposed alternating optimization and co-teaching pipeline is creative and builds sensibly on AlphaFlow and ProteinMPNN; the ablation study and the Spearman-rank analysis are useful sanity checks, and the SARS-CoV-2 RBD case study is a nice qualitative illustration. However, the manuscript's central claim is not currently established: the evaluation is based entirely on Rosetta binding energies computed on computationally predicted and docked structures, and the same energy function is used to train the selection predictor. Without an independent anchor to experimental affinity data, the reported 'state-of-the-art affinity maturation' numbers cannot be distinguished from optimization of a Rosetta proxy. The contribution is therefore best viewed as a computational study of an in silico proxy, and the claims need to be scaled back or externally validated.

major comments (4)
  1. [§3.2, §4.1, Table 1] The central SOTA claim is undermined by a circular evaluation: the sequence-based predictor used to select the top three designs is fine-tuned on Rosetta pairwise ΔΔG labels (§3.2), and the reported IMP is computed with Rosetta ΔG on IgFold-predicted and GeoDock-docked complexes (§4.1). Because the same energy function is both the selection oracle and the ground-truth metric, the high IMP values in Table 1 may simply reflect overfitting to Rosetta's scoring function rather than genuine affinity maturation. The manuscript provides no correlation between Rosetta ΔG and experimentally measured binding affinity and no wet-lab validation. I ask for either an external anchor, such as a Spearman correlation against measured KD or IC50 values on a held-out set, or an explicit reframing of the claim as optimization of the Rosetta proxy, with the 'state-of-the-art affinity maturation' claim removed.
  2. [§4.1, Table 1] Table 1 reports point estimates on only 60 antigens with 3 designs per antigen, and no confidence intervals or significance tests are given. Several headline differences are small (e.g., CDR-H1 IMP: AffinityFlow 88.9% vs AbDPO 88.3%; All-region IMP: 91.2% vs 90.0%), so the claim that AffinityFlow 'consistently achieves the best performance' is not statistically supported. Please report per-antigen variance, bootstrap confidence intervals, or paired significance tests, and state explicitly how many designs are used for each CDR setting.
  3. [§3.1, Eq. (6), §4.3, Appendix C] The guidance relies on the approximation pβ(ΔG|x_t,t) ≈ pβ(ΔG|\hat{x}_1(x_t),1), which the paper itself says is effective only when t is close to 1, and on the additional approximation ∂\hat{x}_1/∂x_t ≈ I in Appendix C. However, the sampling schedule is [1.0, 0.6, 0.3, 0.0], so guidance is applied at t=0.6 and t=0.3 where these approximations are questionable. Please specify at which time steps guidance is actually active, quantify the approximation error, or justify the schedule with an ablation.
  4. [§4.1] The description of the data split is ambiguous with respect to possible leakage. The manuscript says 120 labeled sdAb-antigen pairs are used to initially train predictors, then 77 sdAbs and 54 antigens are extracted to generate 4,158 docked complexes, and the 60 test files have antigens and antibodies not included in the training set. It is not clear whether the 77 sdAbs and 54 antigens used for co-teaching labels are disjoint from the 60 test antibodies and antigens. Please clarify the exact split and confirm that no test antibody or antigen appears in the 4,158 co-teaching complexes.
minor comments (5)
  1. [Table 1, §4.4] The direction of the arrows in the column headers (IMP ↓, Sim ↓, Nat ↓) is confusing because higher IMP and higher Nat are better, while lower Sim is better. Please use explicit arrows or state the orientation in the caption.
  2. [§4.3] The statement 'AlphaFlow sampling steps T to 3 per iteration with a schedule of [1.0, 0.6, 0.3, 0.0]' is ambiguous: with T=3 there should be three intervals, yet four time points are listed. Please clarify the discretization.
  3. [§4.1] The phrase 'resulting in a total of 180 sequence designs' is ambiguous, since 60 antigens × 3 designs equals 180, but the evaluation covers four separate CDR settings; please state whether 180 designs are produced per setting or in total.
  4. [§2.3] The 32-repeat GGGGS linker introduces 160 residues between the antibody and antigen; please discuss whether AlphaFlow, trained on single proteins, can faithfully model such a linked complex and whether this artificial chain affects the validity of the generated structures.
  5. [§4.5] The Spearman correlation analysis is reported on 10 antigens paired with 77 sdAbs, but it is unclear whether these are the same 10 antigens used for the ablation and whether they are held out from co-teaching training; please specify the split and the predictor version used for each reported coefficient.

Circularity Check

2 steps flagged · score 8.0 of 10

The SOTA affinity-maturation claim is scored by the same Rosetta ΔG energy function used to train both predictors and guide sampling, so the reported IMP is largely an optimization of the evaluation metric rather than an independent measurement.

  1. fitted input called prediction [Section 3.2 (Co-teaching) and Section 4.1 (Benchmark/Evaluation)]
    "IMP reflects the proportion of mutated sdAbs with reduced binding energy compared to the original. Structures are predicted using IgFold (Ruffolo et al., 2023), docked with GeoDock (Chu et al., 2023), and binding energies are analyzed via Rosetta (Alford et al., 2017). ... we use Rosetta (Alford et al., 2017) to compute the binding free energy ∆G and then calculate the change in binding free energy ∆∆Gijk = ∆Gij − ∆Gik to form pairwise discrete labels."

    The Rosetta ∆G pipeline is both the source of the labels used to train and refine the sequence- and structure-based predictors and the score that defines the IMP evaluation metric. The paper then 'predicts' affinity maturation by selecting designs with those predictors and evaluating them with the same Rosetta ∆G. A design is counted as improved exactly when its Rosetta ∆G decreases, which is the quantity the predictors were trained to estimate and the guidance was trained to minimize. There is no independent, experiment-derived affinity measurement in the loop; if the Rosetta/GeoDock energies are wrong, the training signal and the benchmark fail together, so the SOTA claim is not independently supported.

  2. fitted input called prediction [Section 3.1, Eq. (8) and Section 4.4 (Results and Analysis)]
    "˜v(xt, t,∆G; θ) = ˆv(xt, t; θ) − γ (1 − t)/t ∇xt ˆfβ(ˆx1(xt)). This vector field guides the ODE sampling process towards lower binding energy. ... The notable IMP is mainly due to our effective predictor guidance, which directs the structure sample generation towards low binding energy."

    Eq. (8) explicitly steers AlphaFlow sampling along the negative gradient of ˆfβ, the structure-based predictor. That predictor is fine-tuned on Rosetta pairwise ∆∆G labels (Section 3.2), and the final sequences are post-selected by the sequence predictor trained on the same labels. The evaluation then reports IMP as the fraction of designs whose Rosetta ∆G is lower than wild type. Thus the generation procedure is constructed to decrease the exact energy function used as the benchmark; the high IMP is a consequence of optimizing the evaluation metric, not evidence of improved binding affinity measured independently.

full rationale

The central circularity is benchmark construction, not formal equation identity. The paper is explicit that Rosetta provides the training labels ('we use Rosetta ... to compute the binding free energy ∆G') and that evaluation is performed with the same tool ('binding energies are analyzed via Rosetta'). Because the structure predictor in Eq. (8) is fine-tuned on those Rosetta labels and the sequence predictor used for post-selection is trained on the same pairwise labels, the method is effectively optimizing the Rosetta score that defines IMP. The paper even attributes the high IMP to 'our effective predictor guidance' that 'directs the structure sample generation towards low binding energy.' This is a fitted-input/called-prediction pattern: the label source and the metric are not independent. No load-bearing self-citation chain was found; AlphaFlow, ProteinMPNN, predictor guidance, and baselines come from external work, so the self-citation patterns do not apply. The score is high because the paper's central state-of-the-art claim rests entirely on this Rosetta-internal loop, with no wet-lab validation or correlation to experimentally measured affinity. The finding is not that the equations are formally tautological; it is that the reported 'prediction' is forced by the choice to train and evaluate on the same computational energy function.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several external tools and domain assumptions. Most importantly, Rosetta Delta G is used both as the label source for predictor training (via the co-teaching module) and as the evaluation metric for IMP, meaning the same energy function appears on both sides of the benchmark. The synthetic all-pairs docking of 77 sdAbs against 54 antigens provides training labels that may not correspond to real complexes. The pretrained AlphaFlow model is used outside its training distribution on concatenated antibody-antigen sequences. These are unverified assumptions that the paper does not test.

free parameters (5)
  • Guidance scaling factor gamma = 5.0
    Tuned on CDR-H3 with 10 antigens; Figure 4 shows IMP depends strongly on it below 5.0.
  • Number of AlphaFlow sampling steps T = 3
    Tuned on CDR-H3; Figure 5 shows IMP requires 3 steps for maximum benefit.
  • Number of alternating optimization iterations = 3
    Chosen to allow up to 9 mutations; no sensitivity analysis reported.
  • Mutation multiplicity schedule = single, double, and triple point mutations per iteration
    Chosen to balance mutation accumulation against structure disruption; no ablation reported.
  • Linker length = 32 GGGGS repeats
    Used to concatenate antibody and antigen sequences for AlphaFlow; no ablations reported.
assumptions (6)
  • domain assumption Rosetta binding free energy Delta G is an adequate proxy for true binding affinity for both training the predictors and evaluating maturation success.
    All predictor labels and all IMP evaluations are computed with Rosetta; no experimental validation is provided, so the central claim depends on this proxy.
  • domain assumption AlphaFlow, trained on single protein chains, can faithfully model the structural ensemble of an antibody-antigen pair concatenated with a 32x GGGGS linker.
    Section 2.3 and Section 4.3: the pretrained model is used without fine-tuning on a linked complex, a setting outside its training distribution.
  • domain assumption GeoDock docking of all 77 sdAbs against all 54 antigens produces physically meaningful complexes with meaningful Rosetta energies.
    Section 3.2: 4,158 docked complexes are generated from 77 times 54 pairs, including combinations that do not naturally bind; this underpins the co-teaching labels.
  • ad hoc to paper The approximation p_beta(Delta G | x_t, t) approximately equals p_beta(Delta G | x_hat_1(x_t), 1) is valid for predictor guidance.
    Section 3.1 Eq. (6); the authors state it is effective only when t is close to 1 and apply guidance in later stages, but the schedule [1.0, 0.6, 0.3, 0.0] with T=3 is not fully specified.
  • domain assumption ESM2-GVP structure-based predictor and ESM-2 sequence-based predictor can be trained to rank antibody-antigen pairs by Delta G with enough accuracy to guide maturation.
    Section 2.4; the authors note the structure-based model may not outperform sequence-based due to unreliable structures. Spearman coefficients on the test set are only 0.51 to 0.52.
  • domain assumption ProteinMPNN inverse folding preserves the structure while introducing mutations that are then scored by the sequence predictor.
    Section 3.1: the generated structure is used as reference for inverse folding; this assumes ProteinMPNN produces sequences compatible with the generated backbone.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AffinityFlow: Guided Flows for Antibody Affinity Maturation." pith.science (2026). https://pith.science/paper/F4PA45T7

@misc{pith2026250210365,
  author       = {Pith},
  title        = {Pith review of: AffinityFlow: Guided Flows for Antibody Affinity Maturation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F4PA45T7}},
  note         = {Machine review of arXiv:2502.10365}
}
read the original abstract

Antibodies are widely used as therapeutics, but their development requires costly affinity maturation, involving iterative mutations to enhance binding affinity.This paper explores a sequence-only scenario for affinity maturation, using solely antibody and antigen sequences. Recently AlphaFlow wraps AlphaFold within flow matching to generate diverse protein structures, enabling a sequence-conditioned generative model of structure. Building on this, we propose an alternating optimization framework that (1) fixes the sequence to guide structure generation toward high binding affinity using a structure-based affinity predictor, then (2) applies inverse folding to create sequence mutations, refined by a sequence-based affinity predictor for post selection. A key challenge is the lack of labeled data for training both predictors. To address this, we develop a co-teaching module that incorporates valuable information from noisy biophysical energies into predictor refinement. The sequence-based predictor selects consensus samples to teach the structure-based predictor, and vice versa. Our method, AffinityFlow, achieves state-of-the-art performance in affinity maturation experiments. We plan to open-source our code after acceptance.

Figures

Figures reproduced from arXiv: 2502.10365 by the authors.

Figure 1
Figure 1. Illustration of alternating optimization. (2) With the high-affinity clean structure, we perform in￾verse folding to introduce targeted mutations, and use a sequence-based predictor for post selection, which identi￾fies promising mutated sequences for the next iteration. A significant challenge in training both predictors is the scarcity of labeled data. To address this, we develop a co￾teaching module that leverage… view at source ↗
Figure 2
Figure 2. Illustration of co-teaching. Using this modeled vector field, we can generate samples from the data distribution pdata(x) by utilizing the corre￾sponding neural Ordinary Differential Equation (ODE). 2.3. AlphaFlow AlphaFold (Jumper et al., 2021) serves as a precise single￾state protein structure predictor, and AlphaFlow (Jing et al., 2024) repurposes AlphaFold within a flow matching frame￾work to generate diverse pr… view at source ↗
Figure 3
Figure 3. Visualizations of model-generated antibody structures bound to the SARS-CoV-2 RBD [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: The three antibody metrics versus scaling [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

77 extracted references · 58 canonical work pages

  1. [1]

    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 format.date year duplicate empty "emp...

  2. [2]

    J., Bambrick, J., et al

    Abramson, J., Adler, J., Dunger, J., Evans, R., Green, T., Pritzel, A., Ronneberger, O., Willmore, L., Ballard, A. J., Bambrick, J., et al. Accurate structure prediction of biomolecular interactions with alphafold 3. Nature, 2024

    work page 2024

  3. [3]

    D., Hu, X., Adachi, Y., Schief, W

    Adolf-Bryfogle, J., Kalyuzhniy, O., Kubitz, M., Weitzner, B. D., Hu, X., Adachi, Y., Schief, W. R., and Dunbrack Jr, R. L. Rosettaantibodydesign (rabd): A general framework for computational antibody design. PLoS computational biology, 2018

    work page 2018

  4. [4]

    A., Smorodina, E., Cotet, T.-S., Flem-Karlsen, K., Frank, R., Mehta, B

    Akbar, R., Bashour, H., Rawat, P., Robert, P. A., Smorodina, E., Cotet, T.-S., Flem-Karlsen, K., Frank, R., Mehta, B. B., Vu, M. H., et al. Progress and challenges for the machine learning-based design of fit-for-purpose monoclonal antibodies. In MAbs. Taylor & Francis, 2022

    work page 2022

  5. [5]

    F., Leaver-Fay, A., Jeliazkov, J

    Alford, R. F., Leaver-Fay, A., Jeliazkov, J. R., OMeara, M. J., DiMaio, F. P., Park, H., Shapovalov, M. V., Renfrew, P. D., Mulligan, V. K., Kappel, K., et al. The rosetta all-atom energy function for macromolecular modeling and design. Journal of chemical theory and computation, 2017

    work page 2017

  6. [6]

    Allgower, E. L. and Georg, K. Numerical continuation methods: an introduction, volume 13. Springer Science & Business Media, 2012

    work page 2012

  7. [7]

    R., Watson, J

    Bennett, N. R., Watson, J. L., Ragotte, R. J., Borst, A. J., See, D. L., Weidle, C., Biswas, R., Shrock, E. L., Leung, P. J., Huang, B., et al. Atomically accurate de novo design of single-domain antibodies. bioRxiv, 2024

    2024

  8. [8]

    and Mitchell, T

    Blum, A. and Mitchell, T. Combining labeled and unlabeled data with co-training. In Proceedings of the eleventh annual conference on Computational learning theory, 1998

    work page 1998

Show all 77 references
  1. [9]

    Pretrainable geometric graph neural network for antibody affinity maturation

    Cai, H., Zhang, Z., Wang, M., Zhong, B., Li, Q., Zhong, Y., Wu, Y., Ying, T., and Tang, J. Pretrainable geometric graph neural network for antibody affinity maturation. Nature Communications, 2024

    2024

  2. [10]

    Generative flows on discrete state-spaces: Enabling multimodal flows with applications to protein co-design

    Campbell, A., Yim, J., Barzilay, R., Rainforth, T., and Jaakkola, T. Generative flows on discrete state-spaces: Enabling multimodal flows with applications to protein co-design. arXiv preprint arXiv:2402.04997, 2024

    2024 arXiv

  3. [11]

    Deep extrapolation for attribute-enhanced generation

    Chan, A., Madani, A., Krause, B., and Naik, N. Deep extrapolation for attribute-enhanced generation. Advances in Neural Information Processing Systems, 34: 0 14084--14096, 2021

    2021

  4. [12]

    S., Liu, H., and Dou, D

    Chen, C., Zheng, S., Chen, X., Dong, E., Liu, X. S., Liu, H., and Dou, D. Generalized dataweighting via class-level gradient manipulation. Advances in Neural Information Processing Systems, 34: 0 14097--14109, 2021

    2021

  5. [13]

    Gradient-based bi-level optimization for deep learning: A survey

    Chen, C., Chen, X., Ma, C., Liu, Z., and Liu, X. Gradient-based bi-level optimization for deep learning: A survey. arXiv preprint arXiv:2207.11719, 2022

    2022 arXiv

  6. [14]

    Bidirectional learning for offline model-based biological sequence design

    Chen, C., Zhang, Y., Liu, X., and Coates, M. Bidirectional learning for offline model-based biological sequence design. In International Conference on Machine Learning. PMLR, 2023 a

    2023

  7. [15]

    Structure-aware protein self-supervised learning

    Chen, C., Zhou, J., Wang, F., Liu, X., and Dou, D. Structure-aware protein self-supervised learning. Bioinformatics, 39 0 (4): 0 btad189, 2023 b

    2023

  8. [16]

    Robust guided diffusion for offline black-box optimization

    Chen, C., Beckham, C., Liu, Z., Liu, X., and Pal, C. Robust guided diffusion for offline black-box optimization. Transactions on Machine Learning Research, 2024 a . ISSN 2835-8856. URL https://openreview.net/forum?id=4JcqmEZ5zt

    2024

  9. [17]

    S., Beckham, C., Liu, Z., Liu, X

    Chen, C. S., Beckham, C., Liu, Z., Liu, X. S., and Pal, C. Parallel-mentoring for offline model-based optimization. Advances in Neural Information Processing Systems, 36, 2024 b

    2024

  10. [18]

    Chiu, M. L. and Gilliland, G. L. Engineering antibody therapeutics. Current opinion in structural biology, 38: 0 163--173, 2016

    2016

  11. [19]

    A., Harmalkar, A., and Gray, J

    Chu, L.-S., Ruffolo, J. A., Harmalkar, A., and Gray, J. J. Flexible protein-protein docking with a multi-track iterative transformer. Protein Science, pp.\ e4862, 2023

    2023

  12. [20]

    J., Milles, L

    Dauparas, J., Anishchenko, I., Bennett, N., Bai, H., Ragotte, R. J., Milles, L. F., Wicky, B. I., Courbet, A., de Haas, R. J., Bethel, N., et al. Robust deep learning--based protein sequence design using proteinmpnn. Science, 2022

    2022

  13. [21]

    and Nichol, A

    Dhariwal, P. and Nichol, A. Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 2021

    2021

  14. [22]

    Dunbar, J., Krawczyk, K., Leem, J., Baker, T., Fuchs, A., Georges, G., Shi, J., and Deane, C. M. Sabdab: the structural antibody database. Nucleic acids research, 2014

    2014

  15. [23]

    Protgpt2 is a deep unsupervised language model for protein design

    Ferruz, N., Schmidt, S., and H \"o cker, B. Protgpt2 is a deep unsupervised language model for protein design. Nature communications, 13 0 (1): 0 4348, 2022

    2022

  16. [24]

    C., Berenberg, D., Zadorozhny, K., Kleinhenz, J., Lafrance-Vanasse, J., Hotzel, I., Wu, Y., Ra, S., Bonneau, R., Cho, K., et al

    Frey, N. C., Berenberg, D., Zadorozhny, K., Kleinhenz, J., Lafrance-Vanasse, J., Hotzel, I., Wu, Y., Ra, S., Bonneau, R., Cho, K., et al. Protein discovery with discrete walk-jump sampling. arXiv preprint arXiv:2306.12360, 2023

    2023 arXiv

  17. [25]

    D., Kosciolek, T., Leman, J

    Gligorijevi \'c , V., Renfrew, P. D., Kosciolek, T., Leman, J. K., Berenberg, D., Vatanen, T., Chandler, C., Taylor, B. C., Fisk, I. M., Vlamakis, H., et al. Structure-based protein function prediction using graph convolutional networks. Nature communications, 2021

    2021

  18. [26]

    T., Sat awa, T., Janusz, B., Shan, L., Zhou, L., R \"o ttger, R., and Krawczyk, K

    Hadsund, J. T., Sat awa, T., Janusz, B., Shan, L., Zhou, L., R \"o ttger, R., and Krawczyk, K. nanobert: a deep learning model for gene agnostic navigation of the nanobody mutational space. Bioinformatics Advances, 2024

    2024

  19. [27]

    Co-teaching: Robust training of deep neural networks with extremely noisy labels

    Han, B., Yao, Q., Yu, X., Niu, G., Xu, M., Hu, W., Tsang, I., and Sugiyama, M. Co-teaching: Robust training of deep neural networks with extremely noisy labels. Advances in neural information processing systems, 31, 2018

    2018

  20. [28]

    J., Oktay, D., Lin, Z., Verkuil, R., Tran, V

    Hayes, T., Rao, R., Akin, H., Sofroniew, N. J., Oktay, D., Lin, Z., Verkuil, R., Tran, V. Q., Deaton, J., Wiggert, M., et al. Simulating 500 million years of evolution with a language model. bioRxiv, pp.\ 2024--07, 2024

    2024

  21. [29]

    L., Shanker, V

    Hie, B. L., Shanker, V. R., Xu, D., Bruun, T. U., Weidenbacher, P. A., Tang, S., Wu, W., Pak, J. E., and Kim, P. S. Efficient evolution of human antibodies from general protein language models. Nature Biotechnology, 2024

    2024

  22. [30]

    Sequence-augmented se (3)-flow matching for conditional protein backbone generation

    Huguet, G., Vuckovic, J., Fatras, K., Thibodeau-Laufer, E., Lemos, P., Islam, R., Liu, C.-H., Rector-Brooks, J., Akhound-Sadegh, T., Bronstein, M., et al. Sequence-augmented se (3)-flow matching for conditional protein backbone generation. arXiv preprint arXiv:2405.20313, 2024

    2024 arXiv

  23. [31]

    Ikram, Z., Liu, D., and Rahman, M. S. Antibody sequence optimization with gradient-guided discrete walk-jump sampling. In ICLR 2024 Workshop on Generative and Experimental Perspectives for Biomolecular Design, 2024

    2024

  24. [32]

    B., Baranov, M., Costello, Z., Barber, K

    Ingraham, J. B., Baranov, M., Costello, Z., Barber, K. W., Wang, W., Ismail, A., Frappier, V., Lord, D. M., Ng-Thow-Hing, C., Van Vlack, E. R., et al. Illuminating protein space with a programmable generative model. Nature, 623, 2023

    2023

  25. [33]

    Jing, B., Eismann, S., Suriana, P., Townshend, R. J. L., and Dror, R. Learning from protein structure with geometric vector perceptrons. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=1YLJDvSx6J4

    2021

  26. [34]

    Eigenfold: Generative protein structure prediction with diffusion models, 2023

    Jing, B., Erives, E., Pao-Huang, P., Corso, G., Berger, B., and Jaakkola, T. Eigenfold: Generative protein structure prediction with diffusion models, 2023

    2023

  27. [35]

    Alphafold meets flow matching for generating protein ensembles, 2024

    Jing, B., Berger, B., and Jaakkola, T. Alphafold meets flow matching for generating protein ensembles, 2024

    2024

  28. [36]

    Highly accurate protein structure prediction with alphafold

    Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Z \' dek, A., Potapenko, A., et al. Highly accurate protein structure prediction with alphafold. nature, 2021

    2021

  29. [37]

    Kingma, D. P. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  30. [38]

    R., Morey-Burrows, F

    Krishna, R., Wang, J., Ahern, W., Sturmfels, P., Venkatesh, P., Kalvet, I., Lee, G. R., Morey-Burrows, F. S., Anishchenko, I., Humphreys, I. R., et al. Generalized biomolecular modeling and design with rosettafold all-atom. Science, 2024

    2024

  31. [39]

    M., and Grover, A

    Krishnamoorthy, S., Mashkaria, S. M., and Grover, A. Diffusion models for black-box optimization. In International Conference on Machine Learning, pp.\ 17842--17857. PMLR, 2023

    2023

  32. [40]

    V., Wang, Y

    Kulyt \.e , P., Vargas, F., Mathis, S. V., Wang, Y. G., Hern \'a ndez-Lobato, J. M., and Li \`o , P. Improving antibody design with force-guided sampling in diffusion models. arXiv preprint arXiv:2406.05832, 2024

    2024 arXiv

  33. [41]

    Voicebox: Text-guided multilingual universal speech generation at scale

    Le, M., Vyas, A., Shi, B., Karrer, B., Sari, L., Moritz, R., Williamson, M., Manohar, V., Adi, Y., Mahadeokar, J., et al. Voicebox: Text-guided multilingual universal speech generation at scale. Advances in neural information processing systems, 2024

    2024

  34. [42]

    Lee, S., Jo, J., and Hwang, S. J. Exploring chemical space with score-based out-of-distribution generation. In International Conference on Machine Learning. PMLR, 2023

    2023

  35. [43]

    S., Farmery, J

    Leem, J., Mitchell, L. S., Farmery, J. H., Barton, J., and Galson, J. D. Deciphering the language of antibodies using self-supervised learning. Patterns, 2022

    2022

  36. [44]

    Full-atom peptide design based on multi-modal flow matching

    Li, J., Cheng, C., Wu, Z., Guo, R., Luo, S., Ren, Z., Peng, J., and Ma, J. Full-atom peptide design based on multi-modal flow matching. arXiv preprint arXiv:2406.00735, 2024

    2024 arXiv

  37. [45]

    A., Lai, Y., Kuo, S.-M., Bao, J., Lan, J., Seeger, M

    Li, T., Cai, H., Yao, H., Zhou, B., Zhang, N., Gong, Y., Zhao, Y., Shen, Q., Qin, W., Hutter, C. A., Lai, Y., Kuo, S.-M., Bao, J., Lan, J., Seeger, M. A., Wong, G., Bi, Y., Lavillette, D., and Li, D. A potent synthetic nanobody targets rbd and protects mice from sars-cov-2 inf...

    2020 doi

  38. [46]

    Lin, H., Wu, L., Huang, Y., Liu, Y., Zhang, O., Zhou, Y., Sun, R., and Li, S. Z. Geoab: Towards realistic antibody design and reliable affinity maturation. bioRxiv, 2024

    2024

  39. [47]

    Language models of protein sequences at the scale of evolution enable accurate structure prediction

    Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., dos Santos Costa, A., Fazel-Zarandi, M., Sercu, T., Candido, S., et al. Language models of protein sequences at the scale of evolution enable accurate structure prediction. BioRxiv, 2022

    2022

  40. [48]

    Evolutionary-scale prediction of atomic-level protein structure with a language model

    Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., Smetanin, N., Verkuil, R., Kabeli, O., Shmueli, Y., et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science, 379, 2023

    2023

  41. [49]

    L., Dror, R

    Lindorff-Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., and Shaw, D. E. Improved side-chain torsion potentials for the amber ff99sb protein force field. Proteins: Structure, Function, and Bioinformatics, 2010

    2010

  42. [50]

    T., Ben-Hamu, H., Nickel, M., and Le, M

    Lipman, Y., Chen, R. T., Ben-Hamu, H., Nickel, M., and Le, M. Flow matching for generative modeling. arXiv preprint arXiv:2210.02747, 2022

    2022 arXiv

  43. [51]

    Deep geometric representations for modeling effects of mutations on protein-protein binding affinity

    Liu, X., Luo, Y., Li, P., Song, S., and Peng, J. Deep geometric representations for modeling effects of mutations on protein-protein binding affinity. PLoS computational biology, 2021

    2021

  44. [52]

    Antigen-specific antibody design and optimization with diffusion-based generative models for protein structures

    Luo, S., Su, Y., Peng, X., Wang, S., Peng, J., and Ma, J. Antigen-specific antibody design and optimization with diffusion-based generative models for protein structures. Advances in Neural Information Processing Systems, 35: 0 9754--9767, 2022

    2022

  45. [53]

    when to update

    Malach, E. and Shalev-Shwartz, S. Decoupling" when to update" from" how to update". Proc. Adv. Neur. Inf. Proc. Syst (NeurIPS), 2017

    2017

  46. [54]

    and Weaver, C

    Murphy, K. and Weaver, C. Janeway's immunobiology. Garland science, 2016

    2016

  47. [55]

    Learning to reweight examples for robust deep learning

    Ren, M., Zeng, W., Yang, B., and Urtasun, R. Learning to reweight examples for robust deep learning. In International conference on machine learning, pp.\ 4334--4343. PMLR, 2018

    2018

  48. [56]

    L., Ma, J., and Fergus, R

    Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C. L., Ma, J., and Fergus, R. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. PNAS, 2019. doi:10.1101/622803. URL https://www....

    2019 doi

  49. [57]

    A., Gray, J

    Ruffolo, J. A., Gray, J. J., and Sulam, J. Deciphering antibody affinity maturation with language models and weakly supervised learning. arXiv preprint arXiv:2112.07782, 2021

    2021 arXiv

  50. [58]

    A., Chu, L.-S., Mahajan, S

    Ruffolo, J. A., Chu, L.-S., Mahajan, S. P., and Gray, J. J. Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies. Nature communications, 2023

    2023

  51. [59]

    Deep learning guided optimization of human antibody against sars-cov-2 variants with broad neutralization

    Shan, S., Luo, S., Yang, Z., Hong, J., Su, Y., Ding, F., Fu, L., Li, C., Chen, P., Ma, J., et al. Deep learning guided optimization of human antibody against sars-cov-2 variants with broad neutralization. Proceedings of the National Academy of Sciences, 2022

    2022

  52. [60]

    W., Ruffolo, J

    Shuai, R. W., Ruffolo, J. A., and Gray, J. J. Generative language modeling for antibody design. bioRxiv, pp.\ 2021--12, 2021

    2021

  53. [61]

    W., Ruffolo, J

    Shuai, R. W., Ruffolo, J. A., and Gray, J. J. Iglm: Infilling language modeling for antibody sequence design. Cell Systems, 2023

    2023

  54. [62]

    P., Kumar, A., Ermon, S., and Poole, B

    Song, Y., Sohl-Dickstein, J., Kingma, D. P., Kumar, A., Ermon, S., and Poole, B. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456, 2020

    2011 arXiv

  55. [63]

    Saprot: Protein language modeling with structure-aware vocabulary

    Su, J., Han, C., Zhou, Y., Shan, J., Zhou, X., and Yuan, F. Saprot: Protein language modeling with structure-aware vocabulary. bioRxiv, pp.\ 2023--10, 2023

    2023

  56. [64]

    Trends in therapeutic antibody affinity maturation: From in-vitro towards next-generation sequencing approaches

    Tabasinezhad, M., Talebkhan, Y., Wenzel, W., Rahimi, H., Omidinia, E., and Mahboudi, F. Trends in therapeutic antibody affinity maturation: From in-vitro towards next-generation sequencing approaches. Immunology letters, 2019

    2019

  57. [65]

    Olsen, I

    Tobias H. Olsen, I. H. M. and Deane, C. M. Ablang: An antibody language model for completing antibody sequences. bioRxiv, 2022. doi:https://doi.org/10.1101/2022.01.20.477061

    2022 doi

  58. [66]

    Victora, G. D. and Nussenzweig, M. C. Germinal centers. Annual review of immunology, 2022

    2022

  59. [67]

    A., Brand, R., Calvo, M

    Wang, Z., Combs, S. A., Brand, R., Calvo, M. R., Xu, P., Price, G., Golovach, N., Salawu, E. O., Wise, C. J., Ponnapalli, S. P., et al. Lm-gvp: an extensible sequence and structure informed deep learning framework for protein property prediction. Scientific reports, 2022

    2022

  60. [68]

    Single domain antibodies: promising experimental and therapeutic tools in infection and immunity

    Wesolowski, J., Alzogaray, V., Reyelt, J., Unger, M., Juarez, K., Urrutia, M., Cauerhff, A., Danquah, W., Rissiek, B., Scheuplein, F., et al. Single domain antibodies: promising experimental and therapeutic tools in infection and immunity. Medical microbiology and immunology, 2009

    2009

  61. [69]

    Bindprofx: assessing mutation-induced binding affinity change by protein interface profiles with pseudo-counts

    Xiong, P., Zhang, C., Zheng, W., and Zhang, Y. Bindprofx: assessing mutation-induced binding affinity change by protein interface profiles with pseudo-counts. Journal of molecular biology, 2017

    2017

  62. [70]

    Peer: a comprehensive and multi-task benchmark for protein sequence understanding

    Xu, M., Zhang, Z., Lu, J., Zhu, Z., Zhang, Y., Chang, M., Liu, R., and Tang, J. Peer: a comprehensive and multi-task benchmark for protein sequence understanding. Advances in Neural Information Processing Systems, 2022

    2022

  63. [71]

    A high-affinity RBD-targeting nanobody improves fusion partner’s potency against SARS-CoV-2

    Yao, H., Cai, H., Li, T., Zhou, B., Qin, W., Lavillette, D., and Li, D. A high-affinity RBD-targeting nanobody improves fusion partner’s potency against SARS-CoV-2 . PLOS Pathogens, 17 0 (3): 0 1--18, 2021. doi:10.1371/journal.ppat.1009328. URL https://doi.org/10.1371/journal....

    2021 doi

  64. [72]

    Proteinbench: A holistic evaluation of protein foundation models

    Ye, F., Zheng, Z., Xue, D., Shen, Y., Wang, L., Ma, Y., Wang, Y., Wang, X., Zhou, X., and Gu, Q. Proteinbench: A holistic evaluation of protein foundation models. arXiv preprint arXiv:2409.06744, 2024

    2024 arXiv

  65. [73]

    S., Liu, Z., Neiswanger, W., and Liu, X

    Yuan, Y., Chen, C. S., Liu, Z., Neiswanger, W., and Liu, X. S. Importance-aware co-teaching for offline model-based optimization. Advances in Neural Information Processing Systems, 36, 2024 a

    2024

  66. [74]

    J., and Liu, X

    Yuan, Y., Zhang, Y., Chen, C., Wu, H., Li, Z., Li, J., Clark, J. J., and Liu, X. Design editing for offline model-based optimization. arXiv preprint arXiv:2405.13964, 2024 b

    2024 arXiv

  67. [75]

    Paretoflow: Guided flows in multi-objective optimization

    Yuan, Y., Chen, C., Pal, C., and Liu, X. Paretoflow: Guided flows in multi-objective optimization. In The Thirteenth International Conference on Learning Representations, 2025. URL https://openreview.net/forum?id=mLyyB4le5u

    2025

  68. [76]

    Zheng, Q., Le, M., Shaul, N., Lipman, Y., Grover, A., and Chen, R. T. Guided flows for generative modeling and decision making. arXiv preprint arXiv:2311.13443, 2023

    2023 arXiv

  69. [77]

    Antigen-specific antibody design via direct energy-based preference optimization

    Zhou, X., Xue, D., Chen, R., Zheng, Z., Wang, L., and Gu, Q. Antigen-specific antibody design via direct energy-based preference optimization. arXiv preprint arXiv:2403.16576, 2024

    2024 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.