Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Conformation-Aware Structure Prediction of Antigen-Recognizing Immune Proteins

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that explicitly labeling structures as apo or holo during training lets one model predict both the unbound and antigen-bound conformations of antibodies, nanobodies, and T-cell receptors from a single sequence.

desk verdict Ibex brings a genuinely new capability—predicting apo and holo antibody structures from one sequence via a conformation token—but the key generalization claim is validated on training pairs, and the abstract oversells state-of-the-art. read the letter →

arxiv 2507.09054 v1 pith:2QCLCO76 submitted 2025-07-11 q-bio.BM cs.LG

Frédéric A. Dreyer , Jan Ludwiczak , Karolis Martinkus , Brennan Abanades , Robert G. Alberstein , Pan Kessel , Pranav Rao , Jae Hyeon Lee
show 3 more authors
Richard Bonneau Andrew M. Watkins Franziska Seeger
This is my paper
classification q-bio.BMcs.LG
keywords antibodystructurepredictionconformationtokenapoholostatesCDRH3loopT-cellreceptormodelingnanobodyout-of-distributiongeneralizationdeeplearningprotein
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

Ibex is a deep-learning model that predicts the three-dimensional structure of the variable domains of antibodies, nanobodies, and T-cell receptors. Its central claim is that explicitly labeling each training structure as apo (unbound) or holo (bound) lets the model learn two distinct, physically meaningful conformations from a single amino-acid sequence, so that at inference a user can ask for either state directly. The paper reports state-of-the-art or competitive accuracy across public benchmarks and on a private set of 286 novel antibodies with unseen CDR H3 loops, while running in under a second per variable domain, far faster than the diffusion-based general predictors it is compared against. If right, this gives therapeutic design a fast way to generate both the unbound and antigen-bound shape of an immune receptor, which matters for docking and for antibodies whose binding follows an induced-fit mechanism.

What carries the argument

The load-bearing component is the conformation token: a one-hot input feature labeling each structure as apo or holo during training, which can be set at inference to request a bound or unbound prediction. Around it, Ibex uses 16 AlphaFold2-style structure module blocks with invariant point attention, a residual connection from the initial embedding to every structure module to preserve the token's influence, ESM-C 300M sequence embeddings, and a three-stage curriculum loss (FAPE, torsion, pLDDT, then structural violation losses) over labeled experimental and distilled data. The final model is an ensemble of eight models returning the prediction closest to the mean.

What would settle it

Take a set of antibody–antigen complexes crystallized both with and without the antigen under identical conditions, ideally newly deposited ones absent from training; if Ibex's apo and holo predictions do not separate into the experimentally observed states, or if shuffling the apo/holo labels during training leaves the two-state prediction gap unchanged, the binding-state conditioning is not capturing real conformational differences.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a structure prediction model can be conditioned on binding state by means of a single conformation token. Trained on structures labeled apo or holo, Ibex produces two accurate, distinct structures from one sequence, recapitulating observed apo/holo conformational differences on 562 matched pairs and predicting hydrogen-bond networks characteristic of each state. The authors argue that previous models, trained on undifferentiated structural databases with multiple entries per sequence, risk predicting a non-physical average or collapsing to the most common state; the conformation token resolves this ambiguity. On the ImmuneBuilder test set Ibex reaches the lowest mean RMSD on TCR CDR beta3 and alpha3 loops and the second lowest on antibody and nanobody CDR H3 loops, and on the private out-of-distribution dataset of 286 novel antibodies it achieves the lowest mean CDR H3 RMSD (2.28 Å) among all compared methods. The authors attribute the out-of-distribution robustness to a three-stage curriculum over experimental immune structures, immunoglobulin-like domains, and a 60k-structure distillation set from predicted models.

Load-bearing premise

The entire apo/holo distinction hangs on the metadata rule that a structure is 'apo' when no antigen chain appears in SAbDab or STCRDab, and 'holo' otherwise; if that binary labeling is noisy, the conformation token learns to separate artifacts rather than true bound and unbound states.

Editorial extensions

If this is right

  • Two structures from one sequence: for any immune receptor sequence, Ibex can output both the apo and holo conformation, so designers can pick the relevant starting state for docking rather than using a single averaged model.
  • State-of-the-art out-of-distribution accuracy on novel CDR H3 loops: on the private dataset of 286 antibodies whose H3 loops differ from all public structures, Ibex's mean CDR H3 RMSD is 2.28 Å, beating both specialized tools like ABodyBuilder3 and general models like Boltz-1 and Boltz-2, suggesting it generalizes to the novel sequences that arise in real therapeutic programs.
  • The model's hydrogen-bond network predictions in the CDRs match the apo/holo ground-truth states in magnitude and connectivity, suggesting the two predicted conformations carry biophysical meaning beyond backbone geometry, with a slight bias toward overproducing holo-state hydrogen bonds.
  • Diffusion-based general predictors (Boltz-1, Chai-1) show almost no improvement in CDR H3 loop accuracy when sampled up to 1000 seeds, whereas Ibex directly conditions on the target state; this argues that for this loop, stochastic sampling does not substitute for explicit binding-state conditioning.
  • Ibex runs in 0.7 s on a single A10G GPU for a variable domain, roughly 10x faster than ESMFold and 90x faster than Boltz-2 including MSA, making high-throughput modeling of immune repertoires practical.

Reading between the lines

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

  • The conformation token is a general architectural trick: since the network already learns to separate conformational states in a single latent space, the same token could be repurposed to condition on other discrete biological states (pH, allosteric ligand binding, oxidation state) without changing the model core, so long as labeled structures exist for those states; the paper suggests the idea bu
  • The apo/holo labeling rule (metadata-only) likely injects label noise: a crystal structure solved without its antigen in the asymmetric unit, or a non-physiological crystal contact, will be mislabeled. The paper's own ablation does not include a label-noise robustness test, so a direct testable extension is to train a version with a fraction of shuffled apo/holo labels and measure how much of the
  • Because the private benchmark set comes from the same industrial pipeline that provided the high-resolution structures, the out-of-distribution numbers are not independently reproducible by outside groups; an independent check on publicly deposited, recently released apo/holo pairs would establish whether the generalization claim holds beyond the training distribution.
  • If Ibex's two predicted states are faithful, the difference vector between apo and holo predictions could serve as a cheap prior for flexible-docking or induced-fit studies, providing starting conformations for refinement without running MD; the paper does not report any docking experiments.
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. Ibex is a deep learning model for structure prediction of antibody, nanobody, and TCR variable domains. It builds on AlphaFold2/ABodyBuilder3 with a conformation token that is set to apo or holo during training and at inference. The model is trained in three stages on SAbDab/STCRDab structures, PDB immunoglobulin-like domains, and ESMFold/Boltz-1 predicted structures from OAS sequences. The paper validates the model on (i) 562 paired apo/holo antibody structures, (ii) the ImmuneBuilder test set, and (iii) a private set of 286 novel antibody structures, and reports a substantial speed advantage over diffusion-based general predictors.

Significance. The strongest contribution is the combination of released inference code and weights with a private, out-of-distribution benchmark of 286 novel antibody structures and an ablation study showing the value of the auxiliary training data. If the conformation-aware claim survives a held-out evaluation, the model would be a useful tool for antibody design. At present, however, the key novelty—predicting distinct and accurate apo and holo structures from a single sequence—is only demonstrated on structures that largely overlap the training set, so its generalization is not yet established.

major comments (4)
  1. [Section 3.1] The apo/holo validation is performed on 562 paired structures, and the paper explicitly states that 'most of the known paired apo/holo structures were included in the training.' Consequently, Figures 2A–C cannot distinguish memorization from generalization, and the central claim that Ibex predicts both conformations for novel sequences is not supported by this analysis. Provide a held-out evaluation, for example by splitting the 562 pairs according to the MMseqs2 cluster definition so that no pair sharing a cluster with a training structure is included, or by repeating the apo/holo analysis on the private dataset of Section 3.3, which per Appendix C contains 177 holo and 109 apo structures. Without such a split, the conformation token may simply implement a learned lookup for training pairs.
  2. [Table 1 / Abstract] The abstract's 'state-of-the-art accuracy' claim is not fully supported by Table 1: Chai-1 achieves a lower mean CDR H3 RMSD than Ibex on antibodies (2.65 Å vs 2.72 Å) and Boltz-1 achieves a lower mean CDR H3 RMSD on nanobodies (2.83 Å vs 3.12 Å). The claim should be qualified to the specific regions and molecule types where Ibex is actually best (e.g., TCR CDR β3 and CDR α3), and the comparison should include measures of variance or paired significance tests, since the reported differences are often small (e.g., 0.02–0.10 Å).
  3. [Section 5.2] The apo/holo label is assigned by the presence of an antigen chain in SAbDab/STCRDab metadata. This binary rule is likely to be noisy: a crystal structure can lack the antigen in the asymmetric unit, and crystallographic contacts can be non-physiological. Because the conformation token is trained directly on these labels, label noise may cause the model to learn spurious distinctions rather than genuine conformational states. The authors should validate a sample of the labels (e.g., manual inspection or comparison with structural metrics such as buried surface area or H3 loop conformation) and analyze the sensitivity of the apo/holo predictions to the labeling rule.
  4. [Section 3.1, Figure 2A] The paper reports a 'reasonable correlation' between predicted and experimental conformational changes but gives no quantitative correlation coefficient, no mean/median loop RMSD values, and no confidence intervals. To support the claim that Ibex recapitulates conformational transitions, report Pearson and Spearman correlations between predicted and observed apo–holo CDR H3 RMSDs, the mean and median signed error, and the fraction of pairs for which the predicted direction of change agrees with experiment.
minor comments (5)
  1. [Section 3.3] The statement that Ibex shows 'comparable performance to Boltz-1' is based on a 0.02 Å difference in mean CDR H3 RMSD (2.28 vs 2.30 Å); report the distribution or a paired test to justify this comparison.
  2. [Section 5.3] The spelling 'SAbdab' appears in the first paragraph of Section 5.3 and should be 'SAbDab' for consistency with the rest of the manuscript.
  3. [Appendix B, Figure 9] The caption of Figure 9 states that the comparison is against ABodyBuilder3, but the surrounding text says the comparison is to TCRBuilder2+; the caption should be corrected.
  4. [Figure 4] The y-axis of Figure 4 is labeled 'Relative improvement' but the plot does not specify whether higher values are better or how the single-seed baseline is defined; please add a precise definition in the caption.
  5. [Section 2] The ensemble procedure 'returns the prediction closest to the mean' should specify the alignment and distance metric used (e.g., global backbone RMSD after superposition on framework residues).

Circularity Check

1 steps flagged · score 6.0 of 10

Apo/holo validation reduces to training-set reconstruction: Section 3.1 evaluates the flagship conformational claim on paired structures that the paper says were mostly in training, and no held-out apo/holo evaluation is provided.

  1. fitted input called prediction [Section 3.1 (Analysis of paired apo/holo structures); training composition in Sections 2, 5.2, 5.3]
    "To validate this, we analyze the conformational changes between 562 experimentally determined apo/holo antibody pairs and compare them to the changes predicted by Ibex. [...] It is important to note here that most of the known paired apo/holo structures were included in the training."

    Section 5.2 reports 760 matched apo/holo pairs within the 14k training structures, and Section 5.3 states these pairs are upsampled and deliberately co-batched during training. The FAPE loss directly supervises backbone coordinates of these structures, so the RMSD values in Figures 2A-C are in-sample reconstruction metrics, not held-out predictions. The paper presents them as validation of the flagship claim that Ibex 'enables accurate prediction of both states at inference time.' No split isolating held-out apo/holo pairs is reported, and the private 286-structure set, although containing 177 holo and 109 apo structures, is not analyzed separately by binding state. Thus the central apo/holo capability claim is supported only by an evaluation that reduces to the training fit.

full rationale

The circular component is confined to the apo/holo capability claim. Section 3.1 evaluates the model on 562 paired apo/holo structures while explicitly noting that most were in training; Sections 5.2 and 5.3 confirm the overlap and show these pairs were upsampled and co-batched. Reporting RMSD on those pairs as 'predictions' measures training-set reconstruction rather than generalization. The rest of the derivation chain is largely self-contained: the ImmuneBuilder benchmark uses clustered-out test structures with leaked nanobody entries removed, and the private benchmark uses novel CDR H3 loops with edit-distance analysis, providing independent support for the general structure-accuracy claims. There is no load-bearing self-citation or invoked uniqueness theorem. The score is 6 rather than higher because the non-circular out-of-distribution benchmarks cover a substantial part of the paper; the circularity is specific to the differentiating apo/holo result, which lacks any held-out evaluation.

Assumptions & free parameters 10 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard deep learning assumptions and on the correctness of the apo/holo labeling and the distillation data. No invented entities are introduced. The model's hyperparameters are hand-chosen and could affect the results, but they are not fitted to the benchmark test sets.

free parameters (10)
  • Stage 1 data source weighting = 30% SAbDab/STCRDab, 40% predicted, 30% Ig-like
    Chosen by hand to balance data sources during pre-training; affects learned representation (Section 5.3).
  • Stage 2 data source weighting = 70% SAbDab/STCRDab, 20% predicted, 10% Ig-like
    Curriculum specialization weighting (Section 5.3).
  • Stage 3 data source weighting = 95% SAbDab/STCRDab, 5% predicted (Boltz-1 only)
    Final fine-tuning weighting (Section 5.3).
  • FAPE loss clamp = 10 A general, 30 A for CDR/framework or loop/non-loop
    Chosen to bound the loss; hyperparameter influencing structure quality (Section 5.3).
  • Apo labeling distance threshold = 10 A
    Heavy atoms within 10 A of a loop residue mark a chain as holo in the immunoglobulin-like dataset; defines binary labels for 22k domains (Section 5.2).
  • CDR H3/beta3 loop length cutoff = 35 residues
    Structures with longer loops are removed, affecting the training distribution (Section 5.2).
  • Abangle outlier cutoff = 5 SD experimental, 3 SD predicted
    Filters structures with unusual VH-VL orientation; hand-chosen thresholds (Section 5.2).
  • Ensemble size = 8
    Eight independently trained models, prediction closest to mean; engineering choice (Sections 3.1, 5.3).
  • Upsampling factors = 1.5x nanobody/TCR, 2x paired apo/holo
    Increases representation of under-sampled categories during training (Section 5.3).
  • Learning rates = 2e-4 (stage 1), 2e-4 to 2e-5 (stage 2), 5e-5 (stage 3)
    Optimization hyperparameters used across the curriculum (Section 5.3).
assumptions (4)
  • domain assumption SAbDab and STCRDab metadata accurately reflects the true binding state of the antibody or TCR in the crystal (apo if no antigen chain is indicated, holo otherwise).
    This labeling rule (Section 5.2) is load-bearing for the apo/holo capability. If the metadata is incomplete or incorrect, the conformation token training labels are noisy.
  • domain assumption Predicted structures from ESMFold and Boltz-1 are sufficiently accurate to serve as additional training data (distillation).
    60k predicted structures are used to improve generalization (Section 5.2). If these predictions are systematically biased, the model inherits that bias.
  • domain assumption The curated set of 562 apo/holo pairs is representative of real conformational changes upon binding.
    Used to validate the apo/holo capability (Section 3.1), though most pairs are in training, which limits the strength of this validation.
  • standard math The AlphaFold2 structure module and OpenFold/ABodyBuilder3 code implementations are correct and reliable.
    Ibex builds on these code bases (Section 5.1); any bug in the underlying implementation would affect all results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Conformation-Aware Structure Prediction of Antigen-Recognizing Immune Proteins." pith.science (2026). https://pith.science/paper/2QCLCO76

@misc{pith2026250709054,
  author       = {Pith},
  title        = {Pith review of: Conformation-Aware Structure Prediction of Antigen-Recognizing Immune Proteins},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2QCLCO76}},
  note         = {Machine review of arXiv:2507.09054}
}
read the original abstract

We introduce Ibex, a pan-immunoglobulin structure prediction model that achieves state-of-the-art accuracy in modeling the variable domains of antibodies, nanobodies, and T-cell receptors. Unlike previous approaches, Ibex explicitly distinguishes between bound and unbound protein conformations by training on labeled apo and holo structural pairs, enabling accurate prediction of both states at inference time. Using a comprehensive private dataset of high-resolution antibody structures, we demonstrate superior out-of-distribution performance compared to existing specialized and general protein structure prediction tools. Ibex combines the accuracy of cutting-edge models with significantly reduced computational requirements, providing a robust foundation for accelerating large molecule design and therapeutic development.

Figures

Figures reproduced from arXiv: 2507.09054 by the authors.

Figure 1
Figure 1. The first two panels show an antibody (PDB codes: 2fr4 and 1xf3), with the holo ground truth and predictions shown on the left, and apo poses shown on the right. The holo Ibex predictions are in dark red and the apo predictions in salmon, superimposed on their respective ground truth structures shown in green and blue. The view is centered on the H3 loop, with side chains shown for the loop residues. The last two pa… view at source ↗
Figure 2
Figure 2. A. RMSD between matching apo and holo pairs, comparing ground truth structures to Ibex predictions. B. RMSD between predicted structures and their corresponding ground truth for matched apo and holo pairs. The mode of the distribution is indicated with a cross. C. Comparison of the number of hydrogen bonds in CDR-CDR residue interaction networks for matching apo and holo pairs. A B C predicted antibody structures. T… view at source ↗
Figure 3
Figure 3. Benchmark on a private dataset of 286 antibody structures, showing the CDR H3 RMSD as a function of edit distance to the closest matching H3 loop in SAbDAb. The box shows the lower and upper quartile, with the median represented as a horizontal line and the mean as a filled circle. Outliers are shown as empty circles, with ESMFold having outliers outside of the plotted range. H3 loop in SAbDab. We observe that while… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Relative improvement in CDR H3 RMSD from consider￾ing the best scoring prediction out of 10, 100 and 1000 seeds for Boltz-1 and Chai-1, evaluated on a private dataset of 286 antibod￾ies and shown as a function of edit distance to the closest matching H3 loop in SAbDab.…
Figure 5
Figure 5. Figure 5: Overview of the Ibex model. The one-hot encoded residue features are concatenated with a projection of language model embeddings and used as input to 16 structure module blocks. The final residue representation is then used to predict all atomic coordinates and uncerta…
Figure 6
Figure 6. Figure 6: Recall (left) and precision (right) of hydrogen bond contacts predicted by Ibex, for pairs of apo/holo structures. 12 [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: CDR and framework RMSD in angstrom, comparing Ibex against ABodyBuilder3 on the ImmuneBuilder test set of 34 antibodies [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: CDR and framework RMSD in angstrom, comparing Ibex against NanoBuilder2 on the ImmuneBuilder test set of 32 nanobodies. The 6 datapoints for which identical CDR H3 sequences were identified in the train or validation split are shown in grey and are excluded from the di…
Figure 9
Figure 9. Figure 9: CDR and framework RMSD in angstrom, comparing Ibex against ABodyBuilder3 on the ImmuneBuilder test set of 21 TCRs. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]
Figure 10
Figure 10. Figure 10: Distribution of CDR H3 edit distance, resolution and CDR H3 length of the private dataset used in Section 3.3. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: CDR H3 RMSD as a function of edit distance to the closest matching H3 loop in SAbDAb, comparing Boltz-2, Boltz-2 without MSA input, and Ibex. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 12
Figure 12. Figure 12: CDR H3 RMSD as a function of edit distance to the closest matching H3 loop in SAbDAb. Left: Impact of the training data size on model performance. We randomly subsample a fraction of training clusters and train a single checkpoint on a reduced dataset of SAbDab and ST…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Tokenizing Loops of Antibodies

    q-bio.BM 2025-09 conditional novelty 6.0 of 10

    Igloo is a multimodal antibody loop tokenizer that, when plugged into protein language models, modestly improves loop retrieval, affinity prediction, and structure-consistent loop generation.

Reference graph

Works this paper leans on

62 extracted references · 50 canonical work pages · cited by 1 Pith paper

  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]

    Abanades, B., Georges, G., Bujotzek, A., and Deane, C. M. Ablooper: fast accurate antibody cdr loop structure prediction with accuracy estimation. Bioinformatics, 38 0 (7): 0 1877--1880, 2022

    work page 2022

  3. [3]

    K., Boyles, F., Georges, G., Bujotzek, A., and Deane, C

    Abanades, B., Wong, W. K., Boyles, F., Georges, G., Bujotzek, A., and Deane, C. M. Immunebuilder: Deep-learning models for predicting the structures of immune proteins. Communications Biology, 6 0 (1): 0 575, 2023

    work page 2023

  4. [4]

    J., Bambrick, J., Bodenstein, S

    Abramson, J., Adler, J., Dunger, J., Evans, R., Green, T., Pritzel, A., Ronneberger, O., Willmore, L., Ballard, A. J., Bambrick, J., Bodenstein, S. W., Evans, D. A., Hung, C.-C., O'Neill, M., Reiman, D., Tunyasuvunakool, K., Wu, Z., Z emgulyt \.e , A., Arvaniti, E., Beattie, C., Bertolli, O., Bridgland, A., Cherepanov, A., Congreve, M., Cowen-Rivers, A. I...

    2024

  5. [5]

    J., Berenberg, D., Fisk, I., Zanichelli, N., et al

    Ahdritz, G., Bouatta, N., Floristean, C., Kadyan, S., Xia, Q., Gerecke, W., O’Donnell, T. J., Berenberg, D., Fisk, I., Zanichelli, N., et al. Openfold: Retraining alphafold2 yields new insights into its learning mechanisms and capacity for generalization. Nature Methods, 21 0 (8): 0 1514--1524, 2024

    work page 2024

  6. [6]

    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, volume 14, pp.\ 2008790. Taylor & Francis, 2022

    work page 2022

  7. [7]

    M., and Chothia, C

    Al-Lazikani, B., Lesk, A. M., and Chothia, C. Standard conformations for the canonical structures of immunoglobulins. Journal of molecular biology, 273 0 (4): 0 927--948, 1997

    work page 1997

  8. [8]

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

    Alford, R. F., Leaver-Fay, A., Jeliazkov, J. R., O’Meara, 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, 13 0 (6): 0 3031--3048, 2017

    work page 2017

Show all 62 references
  1. [9]

    R., Wang, J., Cong, Q., Kinch, L

    Baek, M., DiMaio, F., Anishchenko, I., Dauparas, J., Ovchinnikov, S., Lee, G. R., Wang, J., Cong, Q., Kinch, L. N., Schaeffer, R. D., et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science, 373 0 (6557): 0 871--876, 2021

    2021

  2. [10]

    A structure-based drug discovery paradigm

    Batool, M., Ahmad, B., and Choi, S. A structure-based drug discovery paradigm. International journal of molecular sciences, 20 0 (11): 0 2783, 2019

    2019

  3. [11]

    Curriculum learning

    Bengio, Y., Louradour, J., Collobert, R., and Weston, J. Curriculum learning. In Proceedings of the 26th Annual International Conference on Machine Learning, ICML '09, pp.\ 41–48, New York, NY, USA, 2009. Association for Computing Machinery. ISBN 9781605585161. doi:10.1145/155...

    2009

  4. [12]

    M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T

    Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., and Bourne, P. E. The protein data bank. Nucleic acids research, 28 0 (1): 0 235--242, 2000

    2000

  5. [13]

    C., Melo, M

    Bernardi, R. C., Melo, M. C., and Schulten, K. Enhanced sampling techniques in molecular dynamics simulations of biological systems. Biochimica et Biophysica Acta (BBA)-General Subjects, 1850 0 (5): 0 872--877, 2015

    2015

  6. [14]

    and Thornton, J

    Borkakoti, N. and Thornton, J. M. Alphafold2 protein structure prediction: Implications for drug discovery. Current opinion in structural biology, 78: 0 102526, 2023

    2023

  7. [15]

    J., and Thornton, J

    Borkakoti, N., Ribeiro, A. J., and Thornton, J. M. A structural perspective on enzymes and their catalytic mechanisms. Current Opinion in Structural Biology, 92: 0 103040, 2025

    2025

  8. [16]

    and Porter, L

    Chakravarty, D. and Porter, L. L. Alphafold2 fails to predict protein fold switching. Protein Science, 31 0 (6): 0 e4353, 2022

    2022

  9. [17]

    D., Liao, Y., Kinch, L

    Cheng, H., Schaeffer, R. D., Liao, Y., Kinch, L. N., Pei, J., Shi, S., Kim, B.-H., and Grishin, N. V. Ecod: an evolutionary classification of protein domains. PLoS computational biology, 10 0 (12): 0 e1003926, 2014

    2014

  10. [18]

    Di Noia, J. M. and Neuberger, M. S. Molecular mechanisms of antibody somatic hypermutation. Annu. Rev. Biochem., 76 0 (1): 0 1--22, 2007

    2007

  11. [19]

    C., Roy, A., Maghsoud, Y., Polara, S., Debnath, T., and Cisneros, G

    Dissanayake, U. C., Roy, A., Maghsoud, Y., Polara, S., Debnath, T., and Cisneros, G. A. Computational studies on the functional and structural impact of pathogenic mutations in enzymes. Protein Science, 34 0 (4): 0 e70081, 2025

    2025

  12. [20]

    Dreyer, F. A. Ibex: Conformation-aware structure prediction of antigen-recognizing immune proteins, July 2025. URL https://doi.org/10.5281/zenodo.15866555

    2025 doi

  13. [21]

    and Deane, C

    Dunbar, J. and Deane, C. M. Anarci: antigen receptor numbering and receptor classification. Bioinformatics, 32 0 (2): 0 298--300, 2016

    2016

  14. [22]

    Dunbar, J., Fuchs, A., Shi, J., and Deane, C. M. Abangle: characterising the vh--vl orientation in antibodies. Protein Engineering, Design & Selection, 26 0 (10): 0 611--620, 2013

    2013

  15. [23]

    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, 42 0 (D1): 0 D1140--D1146, 2014

    2014

  16. [24]

    Esm cambrian: Revealing the mysteries of proteins with unsupervised learning, 2024

    ESM Team . Esm cambrian: Revealing the mysteries of proteins with unsupervised learning, 2024. URL https://evolutionaryscale.ai/blog/esm-cambrian

    2024

  17. [25]

    Protein complex prediction with alphafold-multimer

    Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A., Green, T., Z \' dek, A., Bates, R., Blackwell, S., Yim, J., et al. Protein complex prediction with alphafold-multimer. biorxiv, pp.\ 2021--10, 2021

    2021

  18. [26]

    Molecular insight into recognition of the cgrpr complex by migraine prevention therapy aimovig (erenumab)

    Garces, F., Mohr, C., Zhang, L., Huang, C.-S., Chen, Q., King, C., Xu, C., and Wang, Z. Molecular insight into recognition of the cgrpr complex by migraine prevention therapy aimovig (erenumab). Cell Reports, 30 0 (6): 0 1714--1723, 2020

    2020

  19. [27]

    C., Cagiada, M., and Deane, C

    Greenshields-Watson, A., Vavourakis, O., Spoendlin, F. C., Cagiada, M., and Deane, C. M. Challenges and compromises: Predicting unbound antibody structures with deep learning. Current Opinion in Structural Biology, 90: 0 102983, 2025. ISSN 0959-440X. doi:https://doi.org/10.101...

    2025

  20. [28]

    Hitawala, F. N. and Gray, J. J. What does alphafold3 learn about antigen and nanobody docking, and what remains unsolved? bioRxiv, pp.\ 2024--09, 2025

    2024

  21. [29]

    Structure-based drug design with geometric deep learning

    Isert, C., Atz, K., and Schneider, G. Structure-based drug design with geometric deep learning. Current Opinion in Structural Biology, 79: 0 102548, 2023

    2023

  22. [30]

    Alphafold meets flow matching for generating protein ensembles

    Jing, B., Berger, B., and Jaakkola, T. Alphafold meets flow matching for generating protein ensembles. In Forty-first International Conference on Machine Learning, 2024

    2024

  23. [31]

    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, 596 0 (7873): 0 583--589, 2021

    2021

  24. [32]

    and Sander, C

    Kabsch, W. and Sander, C. Dictionary of protein secondary structure: Pattern recognition of hydrogen-bonded and geometrical features. Biopolymers, 22 0 (12): 0 2577--2637, 1983. doi:https://doi.org/10.1002/bip.360221211. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/bip....

    1983 doi

  25. [33]

    Gpu-accelerated homology search with mmseqs2

    Kallenborn, F., Chacon, A., Hundt, C., Sirelkhatim, H., Didi, K., Cha, S., Dallago, C., Mirdita, M., Schmidt, B., and Steinegger, M. Gpu-accelerated homology search with mmseqs2. bioRxiv, pp.\ 2024--11, 2024

    2024

  26. [34]

    and McCammon, J

    Karplus, M. and McCammon, J. A. Molecular dynamics simulations of biomolecules. Nature structural biology, 9 0 (9): 0 646--652, 2002

    2002

  27. [35]

    A., Cutting, D., Nissley, D., and Deane, C

    Kenlay, H., Dreyer, F. A., Cutting, D., Nissley, D., and Deane, C. M. Abodybuilder3: improved and scalable antibody structure predictions. Bioinformatics, 40 0 (10): 0 btae576, 10 2024. ISSN 1367-4811. doi:10.1093/bioinformatics/btae576. URL https://doi.org/10.1093/bioinformat...

    2024 doi

  28. [36]

    M., and Krawczyk, K

    Kovaltsuk, A., Leem, J., Kelm, S., Snowden, J., Deane, C. M., and Krawczyk, K. Observed antibody space: a resource for data mining next-generation sequencing of antibody repertoires. The Journal of Immunology, 201 0 (8): 0 2502--2509, 2018

    2018

  29. [37]

    Pinder: The protein interaction dataset and evaluation resource

    Kovtun, D., Akdel, M., Goncearenco, A., Zhou, G., Holt, G., Baugher, D., Lin, D., Adeshina, Y., Castiglione, T., Wang, X., et al. Pinder: The protein interaction dataset and evaluation resource. bioRxiv, pp.\ 2024--07, 2024

    2024

  30. [38]

    J., Guru, A., Lash, B., Saito, M., Macrae, R

    Kreitz, J., Friedrich, M. J., Guru, A., Lash, B., Saito, M., Macrae, R. K., and Zhang, F. Programmable protein delivery with a bacterial contractile injection system. Nature, 616 0 (7956): 0 357--364, 2023

    2023

  31. [39]

    H., Yadollahpour, P., Watkins, A., Frey, N

    Lee, J. H., Yadollahpour, P., Watkins, A., Frey, N. C., Leaver-Fay, A., Ra, S., Cho, K., Gligorijevi \'c , V., Regev, A., and Bonneau, R. Equifold: Protein structure prediction with a novel coarse-grained structure representation. Biorxiv, pp.\ 2022--10, 2022

    2022

  32. [40]

    Leem, J., de Oliveira, S. H. P., Krawczyk, K., and Deane, C. M. Stcrdab: the structural t-cell receptor database. Nucleic acids research, 46 0 (D1): 0 D406--D412, 2018

    2018

  33. [41]

    Lewis, S., Hempel, T., Jim \'e nez-Luna, J., Gastegger, M., Xie, Y., Foong, A. Y. K., Satorras, V. G., Abdin, O., Veeling, B. S., Zaporozhets, I., Chen, Y., Yang, S., Schneuing, A., Nigam, J., Barbero, F., Stimper, V., Campbell, A., Yim, J., Lienen, M., Shi, Y., Zheng, S., Sch...

    2024 doi

  34. [42]

    A comprehensive review and comparison of existing computational methods for protein function prediction

    Lin, B., Luo, X., Liu, Y., and Jin, X. A comprehensive review and comparison of existing computational methods for protein function prediction. Briefings in Bioinformatics, 25 0 (4): 0 bbae289, 2024

    2024

  35. [43]

    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 0 (6637): 0 1123--1130, 2023

    2023

  36. [44]

    and Hutter, F

    Loshchilov, I. and Hutter, F. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  37. [45]

    Generalizing to any diverse distribution: uniformity, gentle finetuning and rebalancing

    Loukas, A., Martinkus, K., Wagstaff, E., and Cho, K. Generalizing to any diverse distribution: uniformity, gentle finetuning and rebalancing. arXiv preprint arXiv:2410.05980, 2024

    2024 arXiv

  38. [46]

    Development of therapeutic antibodies for the treatment of diseases

    Lu, R.-M., Hwang, Y.-C., Liu, I.-J., Lee, C.-C., Tsai, H.-Z., Li, H.-J., and Wu, H.-C. Development of therapeutic antibodies for the treatment of diseases. Journal of biomedical science, 27: 0 1--30, 2020

    2020

  39. [47]

    Marks, C., Shi, J., and Deane, C. M. Predicting loop conformational ensembles. Bioinformatics, 34 0 (6): 0 949--956, 2018

    2018

  40. [48]

    S., Hopf, T

    Marks, D. S., Hopf, T. A., and Sander, C. Protein structure prediction from sequence variation. Nature biotechnology, 30 0 (11): 0 1072--1080, 2012

    2012

  41. [49]

    Narciso, J. E. T., Uy, I. D. C., Cabang, A. B., Chavez, J. F. C., Pablo, J. L. B., Padilla-Concepcion, G. P., and Padlan, E. A. Analysis of the antibody structure based on high-resolution crystallographic studies. New Biotechnology, 28 0 (5): 0 435--447, 2011. ISSN 1871-6784. ...

    2011 doi

  42. [50]

    North, B., Lehmann, A., and Dunbrack Jr, R. L. A new clustering of antibody cdr loop conformations. Journal of molecular biology, 406 0 (2): 0 228--256, 2011

    2011

  43. [51]

    H., Boyles, F., and Deane, C

    Olsen, T. H., Boyles, F., and Deane, C. M. Observed antibody space: A diverse database of cleaned, annotated, and translated unpaired and paired antibody sequences. Protein Science, 31 0 (1): 0 141--146, 2022

    2022

  44. [52]

    Boltz-2: Towards accurate and efficient binding affinity prediction

    Passaro, S., Corso, G., Wohlwend, J., Reveiz, M., Thaler, S., Ram Somnath, V., Getz, N., Portnoi, T., Roy, J., Stark, H., et al. Boltz-2: Towards accurate and efficient binding affinity prediction. bioRxiv, pp.\ 2025--06, 2025

    2025

  45. [53]

    P., Abanades, B., Guloglu, B., Karuppiah, V., Harper, S., Raybould, M

    Quast, N. P., Abanades, B., Guloglu, B., Karuppiah, V., Harper, S., Raybould, M. I., and Deane, C. M. T-cell receptor structures and predictive models reveal comparable alpha and beta chain structural diversity despite differing genetic complexity. Communications Biology, 8 0 ...

    2025

  46. [54]

    M., Schulze-Gahmen, U., and Wilson, I

    Rini, J. M., Schulze-Gahmen, U., and Wilson, I. A. Structural evidence for induced fit as a mechanism for antibody-antigen recognition. Science, 255 0 (5047): 0 959--965, 1992

    1992

  47. [55]

    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, 14 0 (1): 0 2389, 2023

    2023

  48. [56]

    Schafer, J. W. and Porter, L. L. Alphafold2's training set powers its predictions of some fold-switched conformations. Protein Science, 34 0 (4): 0 e70105, 2025

    2025

  49. [57]

    C., Fern \'a ndez-Quintero, M

    Spoendlin, F. C., Fern \'a ndez-Quintero, M. L., Raghavan, S. S., Turner, H. L., Gharpure, A., Loeffler, J. R., Wong, W. K., Bujotzek, A., Georges, G., Ward, A. B., et al. Predicting the conformational flexibility of antibody and t-cell receptor cdrs. bioRxiv, pp.\ 2025--03, 2025

    2025

  50. [58]

    and S \"o ding, J

    Steinegger, M. and S \"o ding, J. Mmseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nature biotechnology, 35 0 (11): 0 1026--1028, 2017

    2017

  51. [59]

    D., Boitreaud, J., Dent, J., McPartlon, M., Meier, J., Reis, V., Rogozhonikov, A., and Wu, K

    team, C. D., Boitreaud, J., Dent, J., McPartlon, M., Meier, J., Reis, V., Rogozhonikov, A., and Wu, K. Chai-1: Decoding the molecular interactions of life. bioRxiv, 2024. doi:10.1101/2024.10.10.615955. URL https://www.biorxiv.org/content/early/2024/10/11/2024.10.10.615955

    2024 doi

  52. [60]

    and Mizuguchi, K

    Tsuchiya, Y. and Mizuguchi, K. The diversity of h 3 loops determines the antigen-binding tendencies of antibody cdr loops. Protein Science, 25 0 (4): 0 815--825, 2016

    2016

  53. [61]

    Boltz-1 democratizing biomolecular interaction modeling

    Wohlwend, J., Corso, G., Passaro, S., Reveiz, M., Leidal, K., Swiderski, W., Portnoi, T., Chinn, I., Silterra, J., Jaakkola, T., and Barzilay, R. Boltz-1 democratizing biomolecular interaction modeling. bioRxiv, 2024. doi:10.1101/2024.11.19.624167. URL https://www.biorxiv.org/...

    2024 doi

  54. [62]

    Foldbench: An all-atom benchmark for biomolecular structure prediction

    Xu, S., Feng, Q., Qiao, L., Wu, H., Shen, T., Cheng, Y., Zheng, S., and Sun, S. Foldbench: An all-atom benchmark for biomolecular structure prediction. bioRxiv, pp.\ 2025--05, 2025

    2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.