Pith. sign in

REVIEW 5 major objections 4 minor 37 references

CAME-AB: Cross-Modality Attention with Mixture-of-Experts for Antibody Binding Site Prediction

T0 review · 5 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read CAME-AB fuses five modalities of antibody sequence information and reports the best F1, precision, recall, and MCC among four baselines on a seven-class epitope dataset.

desk verdict The forward pass as written leaks the ground-truth label into the fused representation, and the missing inference-time rule makes the reported numbers uninterpretable. read the letter →

arxiv 2509.06465 v4 pith:Z6BYQOCR submitted 2025-09-08 cs.LG cs.CEq-bio.BM

Hongzong Li , Jiahao Ma , Zhanpeng Shi , Rui Xiao , Fanming Jin , Ye-Fan Hu , Hangjun Che , Jian-Dong Huang This is my paper · ORCID
classification cs.LGcs.CEq-bio.BM
keywords antibodybindingsitepredictionantibody-antigeninteractionmultimodallearningadaptivemodalityfusionmixture-of-expertssupervisedcontrastiveproteinlanguagemodelepitopeclassification
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

The paper sets out to show that antibody binding-site prediction is improved by combining several complementary descriptions of the antibody heavy chain: raw residue identity, evolutionary substitution scores, protein-language-model embeddings, predicted structural context, and biochemical similarity graphs. It describes CAME-AB, a network that projects these five views into a shared space, weights them adaptively per sample and per epitope class, and then refines the fused representation with a Transformer and a mixture-of-experts layer. A supervised contrastive loss shapes the latent space so antibodies against the same epitope cluster together. Evaluated on a curated seven-class dataset of antibody sequences, CAME-AB reports top F1 (0.8185), precision (0.8227), recall (0.8250), and MCC (0.7134) against four baselines, with one baseline slightly higher on AUC-ROC. The paper's claim is that the multimodality and the adaptive weighting, not any single feature, are what make the difference.

What carries the argument

The load-bearing mechanism is the adaptive modality fusion module, which computes the fused representation as a weighted sum over modalities, with three kinds of weights: a learnable global importance per modality, a sample-specific gating weight derived from the pooled embedding, and a class-aware weight looked up from the ground-truth epitope class. This fused tensor goes through a two-layer Pre-LN Transformer encoder, whose mean-pooled output is passed to a mixture-of-experts layer with a gating network and a diversity regularizer that penalizes cosine similarity between experts. The final embedding feeds a classifier trained with focal loss plus a supervised contrastive loss, and stochastic weight averaging smooths the last training phase. The design's intent is to let the network keep all five views, but reweight them according to what the input and the target class require.

What would settle it

Re-running the four baselines on the exact train/validation/test split used for CAME-AB, with the same preprocessing and metric definitions, and finding that any baseline reaches or exceeds the reported F1 of 0.8185 or MCC of 0.7134 would falsify the superiority claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a multimodal architecture with adaptive, class-aware fusion can outperform established sequence-based predictors of antibody specificity. The model treats the problem as classification into seven epitope classes (spike subdomains, hemagglutinin head/stem, HIV, and others) from the VH region and the three CDR loops. The authors report that every modality contributes: dropping the ESMC language-model embeddings costs the most F1, removing BLOSUM collapses recall by 17.5 points, and the GCN and one-hot features give smaller but consistent gains. Architecturally, removing the adaptive fusion, the MoE block, contrastive learning, or stochastic weight averaging each lowers F1 or MCC, supporting the claim that the components are complementary. The authors conclude that integrated multimodal representation is a promising direction for immunoinformatics.

Load-bearing premise

The central claim assumes the four baseline scores in Table 3 were measured under the same CD-HIT-clustered 80/10/10 split of the same seven-class curated dataset as CAME-AB, with the same evaluation protocol.

Editorial extensions

If this is right

  • If the results hold, practitioners should include protein-language-model embeddings first: removing them costs the largest F1 drop (about 1.9 percentage points) among the five modalities.
  • The 17.5-point recall drop when BLOSUM is removed implies that evolutionary substitution profiles are what keep the model sensitive across epitope classes, so they should not be treated as redundant with language-model embeddings.
  • Each architectural addition—adaptive fusion, MoE, contrastive learning, and stochastic weight averaging—improves at least one of F1 or MCC, so the gains are attributed to the combination rather than to any single trick.
  • CAME-AB's F1 of 0.8185 and MCC of 0.7134 suggest that class imbalance, which is common in epitope data, is handled well enough by focal loss and contrastive shaping to outperform the four compared baselines on the balanced metrics.

Reading between the lines

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

  • Editorial inference: the paper's title says binding-site prediction, but the experiments classify whole antibody sequences into seven epitope classes; whether the same fusion machinery improves per-residue paratope/epitope annotation remains untested.
  • Editorial inference: Section 4.1.1 does not state whether the four baselines were retrained on the paper's 80/10/10 split or their reported numbers imported from earlier papers, so the reader cannot yet check the fairness of the comparison.
  • Editorial inference: because the class-aware fusion weight is a lookup table over the seven training classes, a model trained this way may not transfer to antibodies against epitopes unseen at training time; a leave-one-epitope-class-out experiment would test that.
  • Editorial inference: the same adaptive fusion and contrastive framework is label-agnostic in design and could be attached to other antibody endpoints, such as neutralization potency or polyreactivity, by only replacing the classification head.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 4 minor

Summary. The paper proposes CAME-AB, a multimodal deep-learning framework for antibody binding site prediction. The model combines five feature modalities derived from antibody VH/CDR sequences: one-hot encoding, BLOSUM profiles, ESMC protein-language-model embeddings, ESMC structure-output features, and GCN-refined biochemical graph embeddings. These are projected into a shared space and fused by an adaptive modality fusion module that uses global, sample-specific, and class-aware weights. The fused representation is processed by a Transformer encoder and a Mixture-of-Experts module, with supervised contrastive learning, focal loss, expert diversity regularization, and stochastic weight averaging. The authors evaluate on a curated antibody-antigen dataset with seven epitope classes and report F1, MCC, precision, recall, and AUC-ROC, claiming state-of-the-art performance. Ablations assess the contribution of each modality and architectural component.

Significance. If the central claims were fully supported, the paper would contribute a useful multimodal architecture for antibody-antigen specificity prediction, and the emphasis on combining pretrained embeddings, structural proxies, and graph features is timely. The availability of code and the use of multiple metrics are also strengths. However, the evaluation as written does not establish the claimed superiority: the class-aware fusion term depends on the ground-truth label with no stated inference-time rule, the headline 'consistently outperforms' is contradicted by the reported AUC-ROC of ME-ACP, and several ablation descriptions do not match the tables. These are load-bearing issues for the paper's main empirical conclusion, not merely presentation problems.

major comments (5)
  1. [Section 3.2.2, Eq. (4)] The fused representation is defined as F_fused = sum_m alpha_m * beta_m^(i) * gamma_m^(y_i) * tilde F^(m), where gamma_m^(y_i) = Embed(y_i)[m] is explicitly a function of the ground-truth epitope class y_i. This makes the label an input feature in the training forward path, and the paper never specifies how gamma is computed at inference. If gamma is omitted at test time, the training and inference distributions differ and the reported test numbers do not describe the deployed model; if gamma is derived from the model's own predicted label, the evaluation is circular. The manuscript must state the inference-time computation of gamma, or remove the class-aware conditioning from the forward path. As written, the reported gains cannot be attributed to multimodal learning rather than label leakage.
  2. [Section 4.3, Table 3, and Abstract] The abstract and Section 4.3 claim CAME-AB 'consistently outperforms' strong baselines, but Table 3 shows ME-ACP achieving AUC-ROC 0.9587 versus CAME-AB's 0.9351. The text dismisses this as 'potential overfitting or miscalibration' without any supporting evidence, such as calibration curves or analysis on validation subsets. The claim should either be weakened to a partial superiority statement or be backed by an explicit calibration comparison. As it stands, the central empirical claim is not supported by the reported numbers.
  3. [Section 4.2.2, Table 2, and surrounding text] The ablation text and Table 2 contain internal inconsistencies. The text states that disabling AMF reduces F1 by 0.92% and MCC by 0.0144, but Table 2 shows a drop of 0.0058 in F1 (0.8185 to 0.8127) and 0.0101 in MCC (0.7134 to 0.7033). The text also states that removing SWA 'leads to the best AUC', yet Table 2 shows that removing the MoE gives AUC 0.9440, which is higher than the w/o SWA value of 0.9397. Additionally, 'removing the MoE block significantly reduces recall' is an overstatement for a drop of 0.0073. These discrepancies need to be corrected before the ablation conclusions can be assessed.
  4. [Section 4.1.1 and Section 4.3] The comparative evaluation is not fully specified. The paper does not report the dataset size, class distribution, or the exact protocol used for the baselines ABS, ME-ACP, xDeep-AcPEP, and PreAlgPro. In particular, it is not stated whether these baselines were retrained on the same 80/10/10 split of the same curated dataset or whether their reported numbers were imported from prior publications. Without this information, Table 3 cannot be interpreted as a fair comparison, and the claim of state-of-the-art performance is not verifiable.
  5. [Section 3.2.4, Eq. (17)] The total loss in Eq. (17) includes the term lambda_aux * L_modal, but L_modal is never defined anywhere in the manuscript or supplementary material. The text refers to 'auxiliary modality losses' but provides no formula or description. This missing definition prevents reproduction of the training objective and should be supplied or the term removed.
minor comments (4)
  1. [Title and Section 3.1] The title promises 'antibody binding site prediction', but the experiments predict one of seven epitope classes (S:NTD, S:RBD, S:S2, Others, HA:Head, HA:Stem, HIV) from antibody VH/CDR sequences. This is better described as antibody specificity or epitope class prediction; the terminology should be aligned to avoid overclaiming.
  2. [Section 3.2.4, Eq. (14)] The cosine similarity function is written as 'com_sim' in the contrastive loss; it should be 'cos_sim' for consistency with Eq. (11) and the surrounding text.
  3. [Section 4.1.2] The grid search over embedding dimension, transformer depth, attention heads, number of experts, contrastive temperature, and loss weights is mentioned, but the final selected values are not reported. Reporting these hyperparameters is necessary for reproducibility.
  4. [Figure 3] The t-SNE plot lacks axis labels that identify which colors correspond to which classes. The caption says the model exhibits class separability, but the reader cannot map the clusters to the seven epitope classes without a legend.

Circularity Check

1 steps flagged · score 6.0 of 10

Eq. (4) feeds the ground-truth label into the fused representation used for classification, so the reported metrics are not shown to be independent predictions from antibody-derived features.

  1. self definitional [Section 3.2.2, Eq. (4); used in forward path Eqs. (6)-(8), (10), (12)]
    "γ(yi)m is a class-aware weight dependent on the ground truth epitope class yi, implemented as a learnable embedding lookup: γ(yi)m = Embed(yi)[m]"

    By Eq. (4), the fused representation F_fused is a function of the ground-truth label y_i through γ. This F_fused is then passed through the Transformer encoder, MoE, and classifier, so the model's prediction is computed from an input that already contains the target label. The paper gives no inference-time rule for γ. If γ is evaluated with the true label, the classifier can read the answer from its input and the reported F1/MCC are not predictions from antibody features. If γ is evaluated with the model's own predicted label, the evaluation is circular by construction. If γ is omitted at test time, training and inference distributions differ, so Table 3 does not describe the deployed model. In every reading, the central superiority claim is not an independent prediction as written.

full rationale

Aside from the label-conditioning in Eq. (4), the paper is largely self-contained: the pretrained ESMC embeddings, PyBioMed descriptors, CD-HIT clustering, and baseline comparisons are external, independent inputs; there is no self-citation chain and no imported uniqueness theorem. Reusing the ABS dataset pipeline is a reproducibility/fairness concern, not circularity. The single load-bearing defect is that the fused representation—and therefore the classifier input—is defined in terms of the ground-truth label, making the reported benchmark metrics potentially constructed from the target itself. Because this affects the paper's strongest empirical claim and no inference-time rule is provided, the circularity score is elevated to 6; the rest of the architecture retains independent content.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard deep learning building blocks, a curated dataset of undisclosed size, and a set of unreported hyperparameters and implementation details. No new physical entities are introduced. The main domain assumptions are that CDR sequences alone determine epitope class and that the ESMC structure head is a valid substitute for experimental structures.

free parameters (5)
  • contrastive temperature tau = not reported
    Temperature in the supervised contrastive loss (Eq. 14) controls embedding geometry; no value is given.
  • loss weights (lambda_contrast, lambda_aux, lambda_div) = not reported
    Weights in Eq. 17 balance focal, auxiliary modality, contrastive, and diversity losses; values are not reported.
  • number of experts K = not reported
    Defines the MoE capacity in Eq. 9-10; no K value is reported.
  • graph edge similarity threshold = not reported
    Controls which residue pairs are connected in the biochemical graph (Section 3.2.1); threshold is unspecified.
  • transformer depth and attention heads = not reported
    Grid searched over depth {2,4,6} and heads {4,8,12}, but the final selected values are not stated.
assumptions (4)
  • domain assumption CDR-H1/H2/H3 sequences of the VH region determine the epitope class
    Section 3.1 formulates epitope prediction as a function of VH/CDR sequences only, ignoring antigen structure entirely.
  • ad hoc to paper ESMC structure output layer provides reliable spatial residue relationships without experimental structures
    Section 3.1 treats this output as a dedicated structural modality, but no validation of this proxy is given in the antibody epitope context.
  • domain assumption 90% CD-HIT clustering and 80% similarity split prevent data leakage
    Section 4.1.1 asserts this stratification guarantees robust generalization, but provides no cluster counts or pairwise identity statistics.
  • standard math Standard deep learning definitions of Transformer, GCN, focal loss, and supervised contrastive loss from cited literature
    Equations 2, 7-8, 14, and 16 are taken directly from prior work and assumed correct.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CAME-AB: Cross-Modality Attention with Mixture-of-Experts for Antibody Binding Site Prediction." pith.science (2026). https://pith.science/paper/Z6BYQOCR

@misc{pith2026250906465,
  author       = {Pith},
  title        = {Pith review of: CAME-AB: Cross-Modality Attention with Mixture-of-Experts for Antibody Binding Site Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z6BYQOCR}},
  note         = {Machine review of arXiv:2509.06465}
}
read the original abstract

Antibody binding site prediction plays a pivotal role in computational immunology and therapeutic antibody design. Existing sequence or structure methods rely on single-view features and fail to identify antibody-specific binding sites on the antigens. In this paper, we propose \textbf{CAME-AB}, a novel Cross-modality Attention framework with a Mixture-of-Experts (MoE) backbone for robust antibody binding site prediction. CAME-AB integrates five biologically grounded modalities, including raw amino acid encodings, BLOSUM substitution profiles, pretrained language model embeddings, structure-aware features, and GCN-refined biochemical graphs, into a unified multimodal representation. To enhance adaptive cross-modal reasoning, we propose an \emph{adaptive modality fusion} module that learns to dynamically weight each modality based on its global relevance and input-specific contribution. A Transformer encoder combined with an MoE module further promotes feature specialization and capacity expansion. We additionally incorporate a supervised contrastive learning objective to explicitly shape the latent space geometry, encouraging intra-class compactness and inter-class separability. To improve optimization stability and generalization, we apply stochastic weight averaging during training. Extensive experiments on benchmark antibody-antigen datasets demonstrate that CAME-AB consistently outperforms strong baselines on multiple metrics, including Precision, Recall, F1-score, AUC-ROC, and MCC. Ablation studies further validate the effectiveness of each architectural component and the benefit of multimodal feature integration. The model implementation details and the codes are available on https://anonymous.4open.science/r/CAME-AB-C525

Figures

Figures reproduced from arXiv: 2509.06465 by the authors.

Figure 1
Figure 1. Comparison between the traditional prediction approach and our proposed cross-modality attention learning [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of our adaptive multimodal transformer for antibody binding site prediction. The model integrates [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. t-SNE visualization of learned feature embeddings. Our model exhibits strong intra-class compactness and [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Precision-Recall curve comparison (micro-averaged). The proposed method maintains consistently high [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 27 canonical work pages

  1. [1]

    Fragment-based computational design of antibodies targeting structured epitopes.Science Advances, 8(45):eabp9540, 2022

    Mauricio Aguilar Rangel, Alice Bedwell, Elisa Costanzi, Ross J Taylor, Rosaria Russo, Gonçalo JL Bernardes, Stefano Ricagno, Judith Frydman, Michele Vendruscolo, and Pietro Sormanni. Fragment-based computational design of antibodies targeting structured epitopes.Science Advances, 8(45):eabp9540, 2022

    work page 2022

  2. [2]

    Epitope- based vaccine design: a comprehensive overview of bioinformatics approaches.Drug Discovery Today, 25(6):1034– 1042, 2020

    Sepideh Parvizpour, Mohammad M Pourseif, Jafar Razmara, Mohammad A Rafi, and Yadollah Omidi. Epitope- based vaccine design: a comprehensive overview of bioinformatics approaches.Drug Discovery Today, 25(6):1034– 1042, 2020

    work page 2020

  3. [3]

    Epitope-focused immunogens targeting the hepatitis c virus glycoproteins induce broadly neutralizing antibodies.Science Advances, 10(49):eado2600, 2024

    Kumar Nagarathinam, Andreas Scheck, Maurice Labuhn, Luisa J Ströh, Elisabeth Herold, Barbora Veselkova, Sarah Tune, Johannes T Cramer, Stéphane Rosset, Sabrina S V ollers, et al. Epitope-focused immunogens targeting the hepatitis c virus glycoproteins induce broadly neutralizing antibodies.Science Advances, 10(49):eado2600, 2024

    work page 2024

  4. [4]

    Potent neutralizing antibodies against multiple epitopes on sars-cov-2 spike.Nature, 584(7821):450–456, 2020

    Lihong Liu, Pengfei Wang, Manoj S Nair, Jian Yu, Micah Rapp, Qian Wang, Yang Luo, Jasper F-W Chan, Vincent Sahi, Amir Figueroa, et al. Potent neutralizing antibodies against multiple epitopes on sars-cov-2 spike.Nature, 584(7821):450–456, 2020. 10 CAME-AB

    work page 2020

  5. [5]

    Challenges in antibody structure prediction

    Monica L Fernández-Quintero, Janik Kokot, Franz Waibl, Anna-Lena M Fischer, Patrick K Quoika, Charlotte M Deane, and Klaus R Liedl. Challenges in antibody structure prediction. InMAbs, volume 15, page 2175319. Taylor & Francis, 2023

    work page 2023

  6. [6]

    Bepipred-3.0: Improved b-cell epitope prediction using protein language models.Protein Science, 31(12):e4497, 2022

    Joakim Nøddeskov Clifford, Magnus Haraldson Høie, Sebastian Deleuran, Bjoern Peters, Morten Nielsen, and Paolo Marcatili. Bepipred-3.0: Improved b-cell epitope prediction using protein language models.Protein Science, 31(12):e4497, 2022

    2022

  7. [7]

    An antibody class with a common cdrh3 motif broadly neutralizes sarbecoviruses.Science Translational Medicine, 14(646):eabn6859, 2022

    Lihong Liu, Sho Iketani, Yicheng Guo, Eswar R Reddem, Ryan G Casner, Manoj S Nair, Jian Yu, Jasper F-W Chan, Maple Wang, Gabriele Cerutti, et al. An antibody class with a common cdrh3 motif broadly neutralizes sarbecoviruses.Science Translational Medicine, 14(646):eabn6859, 2022

    work page 2022

  8. [8]

    Study of basic local alignment search tool (blast) and multiple sequence alignment (clustal-x) of monoclonal mice/human antibodies.BioRxiv, pages 2021–07, 2021

    Ivan Vito Ferrari and Paolo Patrizio. Study of basic local alignment search tool (blast) and multiple sequence alignment (clustal-x) of monoclonal mice/human antibodies.BioRxiv, pages 2021–07, 2021

    work page 2021

Show all 37 references
  1. [9]

    How cryo-electron microscopy and x-ray crystallography complement each other.Protein Science, 26(1):32–39, 2017

    Hong-Wei Wang and Jia-Wei Wang. How cryo-electron microscopy and x-ray crystallography complement each other.Protein Science, 26(1):32–39, 2017

    2017

  2. [10]

    Mmsite: A multi-modal framework for the identification of active sites in proteins.Advances in Neural Information Processing Systems, 37:45819–45849, 2024

    Song Ouyang, Huiyu Cai, Yong Luo, Kehua Su, Lefei Zhang, and Bo Du. Mmsite: A multi-modal framework for the identification of active sites in proteins.Advances in Neural Information Processing Systems, 37:45819–45849, 2024

    2024

  3. [11]

    Learning a generalized graph transformer for protein function prediction in dissimilar sequences.GigaScience, 13:giae093, 2024

    Yiwei Fu, Zhonghui Gu, Xiao Luo, Qirui Guo, Luhua Lai, and Minghua Deng. Learning a generalized graph transformer for protein function prediction in dissimilar sequences.GigaScience, 13:giae093, 2024

    2024

  4. [12]

    Gact-ppis: Prediction of protein-protein interaction sites based on graph structure and transformer network.International Journal of Biological Macromolecules, 283:137272, 2024

    Lu Meng and Huashuai Zhang. Gact-ppis: Prediction of protein-protein interaction sites based on graph structure and transformer network.International Journal of Biological Macromolecules, 283:137272, 2024

    2024

  5. [13]

    Semignn-ppi: Self-ensembling multi-graph neural network for efficient and generalizable protein-protein interaction prediction

    Ziyuan Zhao, Peisheng Qian, Xulei Yang, Zeng Zeng, Cuntai Guan, Wai Leong Tam, and Xiaoli Li. Semignn-ppi: Self-ensembling multi-graph neural network for efficient and generalizable protein-protein interaction prediction. arXiv preprint arXiv:2305.08316, 2023

    2023 arXiv

  6. [14]

    Pf2pi: Protein function prediction based on alphafold2 information and protein-protein interaction

    Ruiqi Li, Peishun Jiao, and Junyi Li. Pf2pi: Protein function prediction based on alphafold2 information and protein-protein interaction. InInternational Conference on Intelligent Computing, pages 278–289. Springer, 2024

    2024

  7. [15]

    Highly accurate protein structure prediction with alphafold.nature, 596(7873):583–589, 2021

    John Jumper, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, Olaf Ronneberger, Kathryn Tunyasuvunakool, Russ Bates, Augustin Žídek, Anna Potapenko, et al. Highly accurate protein structure prediction with alphafold.nature, 596(7873):583–589, 2021

    2021

  8. [16]

    Proteinbert: a universal deep-learning model of protein sequence and function.Bioinformatics, 38(8):2102–2110, 2022

    Nadav Brandes, Dan Ofer, Yam Peleg, Nadav Rappoport, and Michal Linial. Proteinbert: a universal deep-learning model of protein sequence and function.Bioinformatics, 38(8):2102–2110, 2022

    2022

  9. [17]

    Single-residue linear and conformational b cell epitopes prediction using random and esm-2 based projections.Briefings in Bioinformatics, 25(2):bbae084, 2024

    Sapir Israeli and Yoram Louzoun. Single-residue linear and conformational b cell epitopes prediction using random and esm-2 based projections.Briefings in Bioinformatics, 25(2):bbae084, 2024

    2024

  10. [18]

    Replacing the complementarity-determining regions in a human antibody with those from a mouse.Nature, 321(6069):522–525, 1986

    Peter T Jones, Paul H Dear, Jefferson Foote, Michael S Neuberger, and Greg Winter. Replacing the complementarity-determining regions in a human antibody with those from a mouse.Nature, 321(6069):522–525, 1986

    1986

  11. [19]

    Abcpred: a webserver for the discovery of acetyl-and butyryl-cholinesterase inhibitors.Molecular Diversity, pages 1–21, 2022

    Aijaz Ahmad Malik, Suvash Chandra Ojha, Nalini Schaduangrat, and Chanin Nantasenamat. Abcpred: a webserver for the discovery of acetyl-and butyryl-cholinesterase inhibitors.Molecular Diversity, pages 1–21, 2022

    2022

  12. [20]

    Enhanced antibody-antigen structure prediction from molecular docking using alphafold2.Scientific Reports, 13(1):15107, 2023

    Francis Gaudreault, Christopher R Corbeil, and Traian Sulea. Enhanced antibody-antigen structure prediction from molecular docking using alphafold2.Scientific Reports, 13(1):15107, 2023

    2023

  13. [21]

    Using blosum in sequence alignments.Cold Spring Harbor Protocols, 2008(6):pdb–top39, 2008

    David W Mount. Using blosum in sequence alignments.Cold Spring Harbor Protocols, 2008(6):pdb–top39, 2008

    2008

  14. [22]

    Pybiomed: a python library for various molecular representations of chemicals, proteins and dnas and their interactions.Journal of cheminformatics, 10:1–11, 2018

    Jie Dong, Zhi-Jiang Yao, Lin Zhang, Feijun Luo, Qinlu Lin, Ai-Ping Lu, Alex F Chen, and Dong-Sheng Cao. Pybiomed: a python library for various molecular representations of chemicals, proteins and dnas and their interactions.Journal of cheminformatics, 10:1–11, 2018

    2018

  15. [23]

    Semi-supervised classification with graph convolutional networks

    Thomas N Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. In International Conference on Learning Representations, 2017

    2017

  16. [24]

    The devil is in the details: Tackling unimodal spurious correlations for generalizable multimodal reward models.arXiv preprint arXiv:2503.03122, 2025

    Zichao Li, Xueru Wen, Jie Lou, Yuqiu Ji, Yaojie Lu, Xianpei Han, Debing Zhang, and Le Sun. The devil is in the details: Tackling unimodal spurious correlations for generalizable multimodal reward models.arXiv preprint arXiv:2503.03122, 2025

    2025 arXiv

  17. [25]

    On layer normalization in the transformer architecture

    Ruibin Xiong, Yunchang Yang, Di He, Kai Zheng, Shuxin Zheng, Chen Xing, Huishuai Zhang, Yanyan Lan, Liwei Wang, and Tieyan Liu. On layer normalization in the transformer architecture. InInternational Conference on Machine Learning, pages 10524–10533, 2020. 11 CAME-AB

    2020

  18. [26]

    Out- rageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprint arXiv:1701.06538, 2017

    Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. Out- rageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprint arXiv:1701.06538, 2017

    2017 arXiv

  19. [27]

    Diversify and match: A domain adaptive representation learning paradigm for object detection

    Taekyung Kim, Minki Jeong, Seunghyeon Kim, Seokeon Choi, and Changick Kim. Diversify and match: A domain adaptive representation learning paradigm for object detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12456–12465, 2019

    2019

  20. [28]

    Supervised contrastive learning.Advances in Neural Information Processing Systems, 33:18661–18673, 2020

    Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. Supervised contrastive learning.Advances in Neural Information Processing Systems, 33:18661–18673, 2020

    2020

  21. [29]

    Swa object detection.arXiv preprint arXiv:2012.12645, 2020

    Haoyang Zhang, Ying Wang, Feras Dayoub, and Niko Sünderhauf. Swa object detection.arXiv preprint arXiv:2012.12645, 2020

    2012 arXiv

  22. [30]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision, pages 2980–2988, 2017

    2017

  23. [31]

    An explainable language model for antibody specificity prediction using curated influenza hemagglutinin antibodies.Immunity, 57(10):2453–2465, 2024

    Yiquan Wang, Huibin Lv, Qi Wen Teo, Ruipeng Lei, Akshita B Gopal, Wenhao O Ouyang, Yuen-Hei Yeung, Timothy JC Tan, Danbi Choi, Ivana R Shen, et al. An explainable language model for antibody specificity prediction using curated influenza hemagglutinin antibodies.Immunity, 57(1...

    2024

  24. [32]

    Cd-hit: accelerated for clustering the next-generation sequencing data.Bioinformatics, 28(23):3150–3152, 2012

    Limin Fu, Beifang Niu, Zhengwei Zhu, Sitao Wu, and Weizhong Li. Cd-hit: accelerated for clustering the next-generation sequencing data.Bioinformatics, 28(23):3150–3152, 2012

    2012

  25. [33]

    Clustering by pattern similarity in large data sets

    Haixun Wang, Wei Wang, Jiong Yang, and Philip S Yu. Clustering by pattern similarity in large data sets. In Proceedings of the 2002 ACM SIGMOD International Conference on Management of Data, pages 394–405, 2002

    2002

  26. [34]

    Me-acp: multi-view neural networks with ensemble model for identification of anticancer peptides

    Guanwen Feng, Hang Yao, Chaoneng Li, Ruyi Liu, Rungen Huang, Xiaopeng Fan, Ruiquan Ge, and Qiguang Miao. Me-acp: multi-view neural networks with ensemble model for identification of anticancer peptides. Computers in Biology and Medicine, 145:105459, 2022

    2022

  27. [35]

    Jiarui Chen, Hong Hin Cheong, and Shirley WI Siu. xdeep-acpep: deep learning method for anticancer peptide activity prediction based on convolutional neural network and multitask learning.Journal of Chemical Information and Modeling, 61(8):3789–3803, 2021

    2021

  28. [36]

    Lingrong Zhang and Taigang Liu. Prealgpro: Prediction of allergenic proteins with pre-trained protein language model and efficient neutral network.International Journal of Biological Macromolecules, 280:135762, 2024

    2024

  29. [37]

    Prediction of protein solubility based on sequence physicochemical patterns and distributed representation information with deepsolue.BMC biology, 21(1):12, 2023

    Chao Wang and Quan Zou. Prediction of protein solubility based on sequence physicochemical patterns and distributed representation information with deepsolue.BMC biology, 21(1):12, 2023. 12 CAME-AB A Feature Representation in Bioinformatics A.1 ESMC: Evolutionary Substitution ...

    2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.