Pith. sign in

REVIEW 3 major objections 6 minor 62 references

VenusX: Unlocking Fine-Grained Functional Understanding of Proteins

T0 review · 3 major / 6 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 VenusX, the first large-scale benchmark for fine-grained protein function, shows that strong whole-protein performance does not transfer to residue- and fragment-level functional tasks.

desk verdict A well-engineered fine-grained protein function benchmark whose residue-level numbers hinge on an unvalidated negative-label assumption; deserves peer review and a conditional revision. read the letter →

arxiv 2505.11812 v1 pith:JIVPZ47Y submitted 2025-05-17 cs.LG cs.CLq-bio.QM

Yang Tan , Wenrui Gou , Bozitao Zhong , Liang Hong , Huiqun Yu , Bingxin Zhou This is my paper · ORCID
classification cs.LGcs.CLq-bio.QM
keywords proteinfunctionpredictionbenchmarkresidue-levelannotationfragmentclassificationpairwisesimilarityscoringlanguagemodelsstructure-awareout-of-distributiongeneralization
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

VenusX is proposed as the first large-scale benchmark for fine-grained protein functional understanding, spanning residue, fragment, and domain levels with 56 datasets and over 878,000 curated samples. The benchmark's design lets evaluators ask not just whether a model assigns the right global function label, but whether it can locate catalytic sites, binding sites, conserved positions, motifs, domains, and epitopes within a protein. Its baseline results support a central practical message: strong performance on conventional whole-protein tasks does not automatically transfer to fine-grained functional tasks, with cross-family residue prediction of active and binding sites dropping sharply. The authors argue this points to a need for models that capture precise local biological signals rather than global or family-level shortcuts.

What carries the argument

The central object is the benchmark's label construction: curated residue-level annotations from InterPro (active sites, binding sites, conserved sites, motifs, domains), BioLiP (ligand-binding residues from experimental complexes), and SAbDab (epitope residues from antibody-antigen complexes) are turned into binary residue labels, fragment-level family labels, and pair-level same-family labels. The machinery that makes these labels usable is the sequence-identity clustering with MMseqs2 at 50%, 70%, and 90% thresholds, the separate mixed-family and cross-family split protocols that control whether training and test proteins can share families, and the negative-label convention that unannotated residues count as non-functional. Pairwise scoring is driven by the subsampling of 10,000 positive and 10,000 negative pairs per seed, keeping evaluation tractable despite combinatorial pair counts in the hundreds of billions.

What would settle it

Take a random sample of residues VenusX labels negative, check them against experimental mutagenesis or manually curated catalytic-site records, and see whether a nontrivial share are actually functional; if they are, the residue-level tasks reward conservative majority-class predictors rather than genuine functional-site recognition.

Watch

Extended reading notes

Core claim

The paper establishes VenusX as a biologically grounded evaluation platform built from residue-level annotations in InterPro, BioLiP, and SAbDab, processed into three task categories: residue-level binary classification of functional positions, fragment-level multi-class classification into InterPro families, and pairwise functional-similarity scoring without supervision. Across these categories the authors define 56 datasets, with mixed-family and cross-family splits at 50%, 70%, and 90% sequence identity and a pair-subsampling protocol that draws 10,000 positive and 10,000 negative pairs per seed. The documented findings are that protein language models reach high AUPR on in-distribution residue tasks, that sequence-structure hybrids such as SaProt and PROTSSN win most fragment-level and cross-family residue comparisons, that structure aligners such as Foldseek remain the strongest pairwise similarity baseline, and that epitope prediction stays uniformly weak. The paper's interpretive claim is that these patterns show current models often lean on global or distributional cues and do not yet reliably encode localized functional information.

Load-bearing premise

The residue-level tasks assume every unannotated residue is non-functional, so if the source databases miss real functional residues, those residues are scored as negative and every residue-level metric is affected.

Editorial extensions

If this is right

  • Model rankings on fine-grained function will differ from global benchmarks: sequence-structure hybrids like SaProt-650M and PROTSSN lead fragment-level classification and several cross-family residue tasks.
  • Cross-family residue prediction of catalytic and binding sites is far from solved; on Act and BindI the best AUPR drops by roughly 70–80% compared with in-distribution settings.
  • Structure-based aligners remain the reference for pairwise functional similarity: Foldseek achieves near-perfect AUC on some tasks while BLAST trails by more than 40 points, so structure-aware representations matter for detecting functional relatives.
  • Evaluation of protein models should therefore include fine-grained, family-disjoint tasks; global accuracy alone is insufficient evidence of biological understanding.
  • Epitope residue prediction is a consistently weak spot, with no tested model exceeding 0.30 AUPR, marking a concrete gap for antibody and antigen modeling.

Reading between the lines

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

  • Editorial inference: the cross-family AUPR collapse on active and binding sites could serve as a quantitative shortcut-dependence diagnostic, since a model that relies on family-level patterns will drop sharply when test families are unseen.
  • Editorial inference: if the negative-label assumption is incomplete, residue-level rankings may favor models that hedge toward the majority class; masking proteins with sparse annotations would test this.
  • Editorial inference: adding a temporal split—train on InterPro families released before a cutoff, test on families released after—would turn VenusX into a continual-learning benchmark for fine-grained knowledge accumulation.
  • Editorial inference: pair AUC over 10,000 positive and negative pairs mainly measures coarse family separation; stratifying negatives by family distance would probe whether models capture graded functional similarity.
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

3 major / 6 minor

Summary. VenusX is a benchmark for fine-grained protein function understanding at residue, fragment, and protein-pair levels. It curates 7 residue-level binary classification tasks from InterPro, BioLiP, and SAbDab, 5 fragment-level multi-class classification tasks based on InterPro families, and 5 pairwise similarity scoring tasks, together forming 56 datasets with mixed-family and cross-family splits at three sequence identity thresholds. The authors evaluate a broad set of pretrained protein language models, sequence-structure hybrid models, structure-only networks, and alignment-based tools, reporting AUPR, accuracy/Macro-F1, and AUC. The central claims are that VenusX is the first large-scale and biologically grounded fine-grained functional benchmark and that strong performance on global protein-level tasks does not necessarily translate to fine-grained functional understanding.

Significance. If the task-construction concerns are resolved, VenusX would be a valuable resource for evaluating representation models on fine-grained functional annotations. The paper has clear strengths: the curation pipeline is described in detail, split statistics are reported transparently, the training setup is reproducible, and the code and data are publicly released. The baseline coverage is broad for residue-level and fragment-level classification, and the use of AUPR for imbalanced residue tasks is appropriate. The main limitations are that the residue-level labels implicitly treat all unannotated residues as negatives, and the pairwise 'functional similarity' task is defined by InterPro family membership rather than direct functional equivalence. These issues affect the interpretation of the majority of the reported results and the strength of the 'biologically grounded' claim.

major comments (3)
  1. [Section 3.1, Tables 4 and 14-16] Residue-level binary classification implicitly labels every residue not present in the curated annotation as negative. For InterPro, only residues matching annotated signatures are positive; for BioLiP, only residues contacting a co-crystallized ligand are positive; for SAbDab, only residues within the 10 Å Cα cutoff are positive. Because annotation coverage is incomplete and uneven across protein families and sequence regions, many true functional residues are treated as negatives, so AUPR can measure recovery of known annotations rather than biological function. The generic bias statement in Appendix A does not identify this structural assumption, which underpins all seven residue-level tasks and the corresponding cross-family conclusions. Please state this assumption explicitly and add robustness analyses, such as evaluating on high-confidence negatives derived from multiple databases or reporting results stratified by annotation density.
  2. [Section 3.3, Table 6] The pairwise functional similarity task defines ground-truth positive pairs as proteins or fragments belonging to the same InterPro family. InterPro family membership is substantially determined by sequence and structural homology, so this task largely evaluates homology detection rather than direct functional equivalence. The near-perfect AUC values achieved by Foldseek and TM-align are therefore expected under this definition and do not by themselves demonstrate functional understanding. The paper's claim of 'function-based pairing' is stronger than the operational definition supports. Please either reframe this task as family-membership similarity or validate functional relatedness with an independent functional annotation such as shared Gene Ontology terms, and report the agreement between the two.
  3. [Section 3.4, Tables 4-5 and 17] The benchmark description advertises comprehensive evaluation at three sequence identity thresholds and with both mixed-family and cross-family splits for residue-level and fragment-level tasks. The reported baseline experiments, however, cover only a subset: Table 4 reports MF50/MP50/Cross for five InterPro residue targets and MP70/MP90 only for BindP and Epi, while Table 5 and Appendix Table 17 report fragment-level results only for the MF50 split, with no fragment-level cross-family results appearing anywhere. The empirical support for the claims of out-of-distribution and multi-threshold evaluation should be completed, or the claims should be narrowed to match the reported experiments.
minor comments (6)
  1. [Section 4.3 heading] The heading contains a typo: 'Calssification' should be 'Classification'.
  2. [Section 6 heading] The heading 'Disscusion and Conclusion' should be 'Discussion and Conclusion'.
  3. [Tables 2 and 8] Table 2 reports 13,459 classes for the Dom fragment task, while Table 8 reports 12,529 fragment-level and 12,580 protein-level InterPro types for Dom. Please reconcile the discrepancy or explain the filtering difference.
  4. [Section 4.3] The text states that the Dom cross-family AUPR drops by less than 10%, but the best MP50 value is about 0.665 and the best Cross value is about 0.564, which is roughly a 15% relative drop; please clarify whether the comparison is absolute or relative.
  5. [Table 14 and Appendix D] The tables report precision, recall, and F1 for the positive class without stating the decision threshold used to convert model outputs to binary predictions; please specify the thresholding procedure.
  6. [Appendix B.3, Table 9] The cross-family columns in Table 9 are labeled 'Family' and 'Protein', but the caption does not explain that the former counts InterPro families and the latter counts proteins; please make this explicit.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: VenusX is a benchmark-construction paper whose tasks and leaderboard scores are independently grounded in external databases, with no fitted parameter later renamed as a prediction.

full rationale

VenusX does not contain a derivation chain of the kind that can reduce to its own inputs. The benchmark labels are curated from three external resources (InterPro, BioLiP, SAbDab) with explicit, stated criteria for each annotation type. All reported results are evaluations of independently pretrained models on these labels, so the leaderboard numbers are not outputs of a fitted model that was trained on the quantity being 'predicted'. The residue-level tasks implicitly treat unannotated residues as negatives, as described in Section 3.1; this is a label-noise and construct-validity assumption that could bias AUPR scores, but it is not circular because the benchmark does not claim to derive functional truth from the labels. The pairwise similarity tasks define ground truth as 'the pair belongs to the same InterPro family' (Section 3.3); since InterPro families are themselves built partly from sequence and structure homology, this operationalization makes the task partly a homology-retrieval measurement. However, the paper states this definition explicitly and does not claim to have derived functional similarity from first principles; it is a disclosed task definition rather than a circular reduction. Self-citations appear only as tooling (VenusFactory for UniProt extraction), a subsampling strategy reference, and as one of several evaluated baselines (PROTSSN); none is load-bearing for the central benchmark claim. Appendix A discloses generic dataset-bias limitations, which are acknowledged but do not constitute circularity. Overall, the benchmark is self-contained against external databases and benchmarks, so the circularity score is 0.

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

The benchmark introduces no new biological entities or mathematical objects. The free parameters are standard label-definition thresholds and split-construction choices, which are disclosed. The key domain assumptions concern the accuracy and completeness of the source annotations and the use of predicted structures, all of which are reasonable but not independently validated within this paper.

free parameters (4)
  • Binding site distance margin = 0.5 Å
    Used to define BioLiP binding-site residues (Section 2.2). This empirical threshold, added to Van der Waals radii, determines the positive labels for the BindB residue-level task.
  • Epitope Cα distance cutoff = 10 Å
    Used to define SAbDab epitope residues (Section 2.3). This hand-chosen geometric threshold determines all Epi positive labels.
  • Sequence identity thresholds = 50%, 70%, 90%
    Used for MMseqs2 clustering to construct redundancy levels in the splits (Section 3.4). Standard choices, but they define how hard the generalization test is.
  • Pairwise sampling count = 10,000 positive and 10,000 negative pairs per seed
    Subsample size for pairwise scoring evaluation (Section 3.4). Chosen for computational feasibility over the combinatorially large pair space.
assumptions (4)
  • domain assumption Residue-level functional annotations from InterPro, BioLiP, and SAbDab are biologically accurate and complete for the labeled residues.
    The benchmark treats all unlabeled residues as negative for residue-level classification (Section 3.1). Incomplete annotation coverage would introduce false negatives and affect all residue-level tasks.
  • domain assumption InterPro family membership is a valid proxy for functional similarity in the pairwise tasks.
    Positive pairs in the pairwise similarity task are defined by shared InterPro family (Section 3.3). Since InterPro families are largely homology-based, this conflates homology with function.
  • domain assumption AlphaFold predicted structures are accurate enough for structure-based evaluation and structure-based models.
    The benchmark retains only proteins with AlphaFold structures and uses them for GVP-GNN and structure alignment baselines (Section 2.1, Appendix B.3). Errors in predicted structures propagate to structure-based scores.
  • domain assumption MMseqs2 clustering at the chosen identity thresholds and coverage 0.8 produces non-redundant splits.
    Split construction relies on MMseqs2 (Section 3.4). The coverage and identity thresholds affect how much redundancy remains between train and test sets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VenusX: Unlocking Fine-Grained Functional Understanding of Proteins." pith.science (2026). https://pith.science/paper/JIVPZ47Y

@misc{pith2026250511812,
  author       = {Pith},
  title        = {Pith review of: VenusX: Unlocking Fine-Grained Functional Understanding of Proteins},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JIVPZ47Y}},
  note         = {Machine review of arXiv:2505.11812}
}
read the original abstract

Deep learning models have driven significant progress in predicting protein function and interactions at the protein level. While these advancements have been invaluable for many biological applications such as enzyme engineering and function annotation, a more detailed perspective is essential for understanding protein functional mechanisms and evaluating the biological knowledge captured by models. To address this demand, we introduce VenusX, the first large-scale benchmark for fine-grained functional annotation and function-based protein pairing at the residue, fragment, and domain levels. VenusX comprises three major task categories across six types of annotations, including residue-level binary classification, fragment-level multi-class classification, and pairwise functional similarity scoring for identifying critical active sites, binding sites, conserved sites, motifs, domains, and epitopes. The benchmark features over 878,000 samples curated from major open-source databases such as InterPro, BioLiP, and SAbDab. By providing mixed-family and cross-family splits at three sequence identity thresholds, our benchmark enables a comprehensive assessment of model performance on both in-distribution and out-of-distribution scenarios. For baseline evaluation, we assess a diverse set of popular and open-source models, including pre-trained protein language models, sequence-structure hybrids, structure-based methods, and alignment-based techniques. Their performance is reported across all benchmark datasets and evaluation settings using multiple metrics, offering a thorough comparison and a strong foundation for future research. Code and data are publicly available at https://github.com/ai4protein/VenusX.

Figures

Figures reproduced from arXiv: 2505.11812 by the authors.

Figure 1
Figure 1. Overview of the VENUSX benchmark. (a) Six types of functional annotations collected from InterPro, BioLiP, and SAbDab (Section 2). (b) Three benchmark task categories: residue-level and fragment-level classification, and pairwise similarity scoring (Sections 3.1-3.3). (c) Sequence identity–based clustering and mix-family and cross-family data split strategies (Sections 3.4). To ensure non-redundancy, we remove addit… view at source ↗
Figure 2
Figure 2. Sequence length distribution of the VENUSX InterPro benchmark [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. InterPro label distribution of the VENUSX benchmark [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 46 canonical work pages

  1. [1]

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

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

    2021

  2. [2]

    Accurate structure prediction of biomolecular interactions with AlphaFold 3.Nature, pages 1–3, 2024

    Josh Abramson, Jonas Adler, Jack Dunger, Richard Evans, Tim Green, Alexander Pritzel, Olaf Ronneberger, Lindsay Willmore, Andrew J Ballard, Joshua Bambrick, et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3.Nature, pages 1–3, 2024

    2024

  3. [3]

    Machine learning- aided engineering of hydrolases for PET depolymerization.Nature, 604(7907):662–667, 2022

    Hongyuan Lu, Daniel J Diaz, Natalie J Czarnecki, Congzhi Zhu, Wantae Kim, Raghav Shroff, Daniel J Acosta, Bradley R Alexander, Hannah O Cole, Yan Zhang, et al. Machine learning- aided engineering of hydrolases for PET depolymerization.Nature, 604(7907):662–667, 2022

    work page 2022

  4. [4]

    A conditional protein diffusion model generates artificial programmable endonuclease sequences with enhanced activity.Cell Discovery, 10(1):95, 2024

    Bingxin Zhou, Lirong Zheng, Banghao Wu, Kai Yi, Bozitao Zhong, Yang Tan, Qian Liu, Pietro Liò, and Liang Hong. A conditional protein diffusion model generates artificial programmable endonuclease sequences with enhanced activity.Cell Discovery, 10(1):95, 2024

    work page 2024

  5. [5]

    Retrieval-enhanced mu- tation mastery: Augmenting zero-shot prediction of protein language model.arXiv:2410.21127, 2024

    Yang Tan, Ruilin Wang, Banghao Wu, Liang Hong, and Bingxin Zhou. Retrieval-enhanced mu- tation mastery: Augmenting zero-shot prediction of protein language model.arXiv:2410.21127, 2024

    arXiv 2024

  6. [6]

    Enzyme function prediction using contrastive learning.Science, 379(6639):1358–1363, 2023

    Tianhao Yu, Haiyang Cui, Jianan Canal Li, Yunan Luo, Guangde Jiang, and Huimin Zhao. Enzyme function prediction using contrastive learning.Science, 379(6639):1358–1363, 2023

    2023

  7. [7]

    Enhancing efficiency of protein language models with minimal wet-lab data through few-shot learning.Nature Communications, 15(1):5566, 2024

    Ziyi Zhou, Liang Zhang, Yuanxi Yu, Banghao Wu, Mingchen Li, Liang Hong, and Pan Tan. Enhancing efficiency of protein language models with minimal wet-lab data through few-shot learning.Nature Communications, 15(1):5566, 2024

    work page 2024

  8. [8]

    CATH – a hierarchic classification of protein domain structures.Structure, 5(8):1093–1109, 1997

    CA Orengo, AD Michie, S Jones, DT Jones, MB Swindells, and JM Thornton. CATH – a hierarchic classification of protein domain structures.Structure, 5(8):1093–1109, 1997

    work page 1997

Show all 62 references
  1. [9]

    Alphafold protein structure database: massively expanding the structural coverage of protein-sequence space with high-accuracy models.Nucleic Acids Research, 50(D1):D439–D444, 2022

    Mihaly Varadi, Stephen Anyango, Mandar Deshpande, Sreenath Nair, Cindy Natassia, Galabina Yordanova, David Yuan, Oana Stroe, Gemma Wood, Agata Laydon, et al. Alphafold protein structure database: massively expanding the structural coverage of protein-sequence space with high-a...

    2022

  2. [10]

    UniProt: the universal protein knowledgebase in 2025.Nucleic Acids Research, 53(D1):D609–D617, 2025

    UniProt Consortium. UniProt: the universal protein knowledgebase in 2025.Nucleic Acids Research, 53(D1):D609–D617, 2025

    2025

  3. [11]

    Simple, efficient, and scalable structure-aware adapter boosts protein language models.Journal of Chemical Information and Modeling, 2024

    Yang Tan, Mingchen Li, Bingxin Zhou, Bozitao Zhong, Lirong Zheng, Pan Tan, Ziyi Zhou, Huiqun Yu, Guisheng Fan, and Liang Hong. Simple, efficient, and scalable structure-aware adapter boosts protein language models.Journal of Chemical Information and Modeling, 2024

    2024

  4. [12]

    Immunogenicity prediction with dual attention enables vaccine target selection

    Song Li, Yang Tan, Song Ke, Liang Hong, and Bingxin Zhou. Immunogenicity prediction with dual attention enables vaccine target selection. InThe Thirteenth International Conference on Learning Representations, 2025

    2025

  5. [13]

    Large-scale prediction of human protein- protein interactions from amino acid sequence based on latent topic features.Journal of Proteome Research, 9(10):4992–5001, 2010

    Xiao-Yong Pan, Ya-Nan Zhang, and Hong-Bin Shen. Large-scale prediction of human protein- protein interactions from amino acid sequence based on latent topic features.Journal of Proteome Research, 9(10):4992–5001, 2010

    2010

  6. [14]

    Damian Szklarczyk, Annika L Gable, David Lyon, Alexander Junge, Stefan Wyder, Jaime Huerta-Cepas, Milan Simonovic, Nadezhda T Doncheva, John H Morris, Peer Bork, et al. String v11: protein–protein association networks with increased coverage, supporting functional discovery in...

    2019

  7. [15]

    Skempi 2.0: an updated benchmark of changes in protein–protein binding energy, kinetics and thermodynamics upon mutation.Bioinformatics, 35(3):462–469, 2019

    Justina Jankauskait˙e, Brian Jiménez-García, Justas Dapk¯unas, Juan Fernández-Recio, and Iain H Moal. Skempi 2.0: an updated benchmark of changes in protein–protein binding energy, kinetics and thermodynamics upon mutation.Bioinformatics, 35(3):462–469, 2019

    2019

  8. [16]

    A comprehensive dataset of protein-protein interactions and ligand binding pockets for advancing drug discovery.Scientific Data, 11(1):402, 2024

    Alexandra Moine-Franel, Fabien Mareuil, Michael Nilges, Constantin Bogdan Ciambur, and Olivier Sperandio. A comprehensive dataset of protein-protein interactions and ligand binding pockets for advancing drug discovery.Scientific Data, 11(1):402, 2024. 10

    2024

  9. [17]

    Quan- titative missense variant effect prediction using large-scale mutagenesis data.Cell Systems, 6(1):116–124, 2018

    Vanessa E Gray, Ronald J Hause, Jens Luebeck, Jay Shendure, and Douglas M Fowler. Quan- titative missense variant effect prediction using large-scale mutagenesis data.Cell Systems, 6(1):116–124, 2018

    2018

  10. [18]

    Deep generative models of genetic variation capture the effects of mutations.Nature Methods, 15(10):816–822, 2018

    Adam J Riesselman, John B Ingraham, and Debora S Marks. Deep generative models of genetic variation capture the effects of mutations.Nature Methods, 15(10):816–822, 2018

    2018

  11. [19]

    ProteinGym: large-scale benchmarks for protein fitness prediction and design

    Pascal Notin, Aaron Kollasch, Daniel Ritter, Lood Van Niekerk, Steffanie Paul, Han Spinner, Nathan Rollins, Ada Shaw, Rose Orenbuch, Ruben Weitzman, et al. ProteinGym: large-scale benchmarks for protein fitness prediction and design. InAdvances in Neural Information Processing...

    2024

  12. [20]

    Venusmuthub: a systematic evaluation of protein mutation effect predictors on small-scale experimental data.Acta Pharmaceutica Sinica B, 2025

    Liang Zhang, Hua Pang, Chenghao Zhang, Song Li, Yang Tan, Fan Jiang, Mingchen Li, Yuanxi Yu, Ziyi Zhou, Banghao Wu, et al. Venusmuthub: a systematic evaluation of protein mutation effect predictors on small-scale experimental data.Acta Pharmaceutica Sinica B, 2025

    2025

  13. [21]

    Discovering functionally important sites in proteins.Nature communications, 14(1):4175, 2023

    Matteo Cagiada, Sandro Bottaro, Søren Lindemose, Signe M Schenstrøm, Amelie Stein, Rasmus Hartmann-Petersen, and Kresten Lindorff-Larsen. Discovering functionally important sites in proteins.Nature communications, 14(1):4175, 2023

    2023

  14. [22]

    Predicting protein function from sequence and structure.Nature reviews molecular cell biology, 8(12):995–1005, 2007

    David Lee, Oliver Redfern, and Christine Orengo. Predicting protein function from sequence and structure.Nature reviews molecular cell biology, 8(12):995–1005, 2007

    2007

  15. [23]

    Deep- bce: evaluation of deep learning models for identification of immunogenic b-cell epitopes

    Muhammad Attique, Tamim Alkhalifah, Fahad Alturise, and Yaser Daanial Khan. Deep- bce: evaluation of deep learning models for identification of immunogenic b-cell epitopes. Computational Biology and Chemistry, 104:107874, 2023

    2023

  16. [24]

    InterPro in 2022.Nucleic Acids Research, 51(D1):D418–D427, 2023

    Typhaine Paysan-Lafosse, Matthias Blum, Sara Chuguransky, Tiago Grego, Beatriz Lázaro Pinto, Gustavo A Salazar, Maxwell L Bileschi, Peer Bork, Alan Bridge, Lucy Colwell, et al. InterPro in 2022.Nucleic Acids Research, 51(D1):D418–D427, 2023

    2022

  17. [25]

    BioLiP: a semi-manually curated database for biologically relevant ligand–protein interactions.Nucleic Acids Research, 41(D1):D1096– D1103, 2012

    Jianyi Yang, Ambrish Roy, and Yang Zhang. BioLiP: a semi-manually curated database for biologically relevant ligand–protein interactions.Nucleic Acids Research, 41(D1):D1096– D1103, 2012

    2012

  18. [26]

    SAbDab: the structural antibody database.Nucleic Acids Research, 42(D1):D1140–D1146, 2014

    James Dunbar, Konrad Krawczyk, Jinwoo Leem, Terry Baker, Angelika Fuchs, Guy Georges, Jiye Shi, and Charlotte M Deane. SAbDab: the structural antibody database.Nucleic Acids Research, 42(D1):D1140–D1146, 2014

    2014

  19. [27]

    Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C Lawrence Zitnick, Jerry Ma, et al. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences.Proceedings of the National A...

    2021

  20. [28]

    Prottrans: Toward understanding the language of life through self-supervised learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(10):7112–7127, 2021

    Ahmed Elnaggar, Michael Heinzinger, Christian Dallago, Ghalia Rehawi, Yu Wang, Llion Jones, Tom Gibbs, Tamas Feher, Christoph Angerer, Martin Steinegger, et al. Prottrans: Toward understanding the language of life through self-supervised learning.IEEE Transactions on Pattern A...

    2021

  21. [29]

    Ankh: Optimized protein language model unlocks general-purpose modelling.arXiv:2301.06568, 2023

    Ahmed Elnaggar, Hazem Essam, Wafaa Salah-Eldin, Walid Moustafa, Mohamed Elkerdawy, Charlotte Rochereau, and Burkhard Rost. Ankh: Optimized protein language model unlocks general-purpose modelling.arXiv:2301.06568, 2023

    2023 arXiv

  22. [30]

    ProstT5: Bilingual language model for protein sequence and structure.bioRxiv, pages 2023–07, 2023

    Michael Heinzinger, Konstantin Weissenow, Joaquin Gomez Sanchez, Adrian Henkel, Martin Steinegger, and Burkhard Rost. ProstT5: Bilingual language model for protein sequence and structure.bioRxiv, pages 2023–07, 2023

    2023

  23. [31]

    Evolutionary-scale prediction of atomic-level protein structure with a language model.Science, 379(6637):1123–1130, 2023

    Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Nikita Smetanin, Robert Verkuil, Ori Kabeli, Yaniv Shmueli, et al. Evolutionary-scale prediction of atomic-level protein structure with a language model.Science, 379(6637):1123–1130, 2023

    2023

  24. [32]

    Protein remote homology detection and structural alignment using deep learning

    Tymor Hamamsy, James T Morton, Robert Blackwell, Daniel Berenberg, Nicholas Carriero, Vladimir Gligorijevic, Charlie EM Strauss, Julia Koehler Leman, Kyunghyun Cho, and Richard Bonneau. Protein remote homology detection and structural alignment using deep learning. Nature Biot...

    2024

  25. [33]

    SaProt: protein language modeling with structure-aware vocabulary

    Jin Su, Chenchen Han, Yuyang Zhou, Junjie Shan, Xibin Zhou, and Fajie Yuan. SaProt: protein language modeling with structure-aware vocabulary. InThe Twelfth International Conference on Learning Representations, 2023

    2023

  26. [34]

    Semantical and geo- metrical protein encoding toward enhanced bioactivity and thermostability.eLife, 13:RP98033, may 2025

    Yang Tan, Bingxin Zhou, Lirong Zheng, Guisheng Fan, and Liang Hong. Semantical and geo- metrical protein encoding toward enhanced bioactivity and thermostability.eLife, 13:RP98033, may 2025

    2025

  27. [35]

    Masked inverse folding with sequence transfer for protein representation learning.Protein Engineering, Design and Selection, 36:gzad015, 2023

    Kevin K Yang, Niccolò Zanichelli, and Hugh Yeh. Masked inverse folding with sequence transfer for protein representation learning.Protein Engineering, Design and Selection, 36:gzad015, 2023

    2023

  28. [36]

    Learning inverse folding from millions of predicted structures

    Chloe Hsu, Robert Verkuil, Jason Liu, Zeming Lin, Brian Hie, Tom Sercu, Adam Lerer, and Alexander Rives. Learning inverse folding from millions of predicted structures. InInternational Conference on Machine Learning, pages 8946–8970. PMLR, 2022

    2022

  29. [37]

    Learning from protein structure with geometric vector perceptrons

    Bowen Jing, Stephan Eismann, Patricia Suriana, Raphael John Lamarre Townshend, and Ron Dror. Learning from protein structure with geometric vector perceptrons. InInternational Conference on Learning Representations, 2021

    2021

  30. [38]

    Basic local alignment search tool.Journal of Molecular Biology, 215(3):403–410, 1990

    Stephen F Altschul, Warren Gish, Webb Miller, Eugene W Myers, and David J Lipman. Basic local alignment search tool.Journal of Molecular Biology, 215(3):403–410, 1990

    1990

  31. [39]

    TM-align: a protein structure alignment algorithm based on the tm-score.Nucleic Acids Research, 33(7):2302–2309, 2005

    Yang Zhang and Jeffrey Skolnick. TM-align: a protein structure alignment algorithm based on the tm-score.Nucleic Acids Research, 33(7):2302–2309, 2005

    2005

  32. [40]

    Fast and accurate protein structure search with Foldseek.Nature Biotechnology, 42(2):243–246, 2024

    Michel Van Kempen, Stephanie S Kim, Charlotte Tumescheit, Milot Mirdita, Jeongjae Lee, Cameron LM Gilchrist, Johannes Söding, and Martin Steinegger. Fast and accurate protein structure search with Foldseek.Nature Biotechnology, 42(2):243–246, 2024

    2024

  33. [41]

    Venusfactory: A unified platform for protein engineering data retrieval and language model fine-tuning.arXiv:2503.15438, 2025

    Yang Tan, Chen Liu, Jingyuan Gao, Banghao Wu, Mingchen Li, Ruilin Wang, Lingrong Zhang, Huiqun Yu, Guisheng Fan, Liang Hong, and Bingxin Zhou. Venusfactory: A unified platform for protein engineering data retrieval and language model fine-tuning.arXiv:2503.15438, 2025

    2025 arXiv

  34. [42]

    Stephen K Burley, Helen M Berman, Charmi Bhikadiya, Chunxiao Bi, Li Chen, Luigi Di Costanzo, Cole Christie, Ken Dalenberg, Jose M Duarte, Shuchismita Dutta, et al. RCSB Protein Data Bank: biological macromolecular structures enabling research and education in fundamental biolo...

    2019

  35. [43]

    BioPython: freely available python tools for computational molecular biology and bioinformatics.Bioinformatics, 25(11):1422, 2009

    Peter JA Cock, Tiago Antao, Jeffrey T Chang, Brad A Chapman, Cymon J Cox, Andrew Dalke, Iddo Friedberg, Thomas Hamelryck, Frank Kauff, Bartek Wilczynski, et al. BioPython: freely available python tools for computational molecular biology and bioinformatics.Bioinformatics, 25(1...

    2009

  36. [44]

    MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets.Nature Biotechnology, 35(11):1026–1028, 2017

    Martin Steinegger and Johannes Söding. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets.Nature Biotechnology, 35(11):1026–1028, 2017

    2017

  37. [45]

    Yang Tan, Lirong Zheng, Bozitao Zhong, Liang Hong, and Bingxin Zhou. Protein representation learning with sequence information embedding: Does it always lead to a better performance? In2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pages 233–239. ...

    2024

  38. [46]

    Fixing weight decay regularization in adam

    Ilya Loshchilov, Frank Hutter, et al. Fixing weight decay regularization in adam. arXiv:1711.05101, 5, 2017

    2017 arXiv

  39. [47]

    Evaluating protein transfer learning with TAPE.Advances in Neural Information Processing Systems, 32, 2019

    Roshan Rao, Nicholas Bhattacharya, Neil Thomas, Yan Duan, Peter Chen, John Canny, Pieter Abbeel, and Yun Song. Evaluating protein transfer learning with TAPE.Advances in Neural Information Processing Systems, 32, 2019

    2019

  40. [48]

    ProteinNet: a standardized data set for machine learning of protein structure.BMC Bioinformatics, 20:1–10, 2019

    Mohammed AlQuraishi. ProteinNet: a standardized data set for machine learning of protein structure.BMC Bioinformatics, 20:1–10, 2019. 12

    2019

  41. [49]

    PEER: A comprehensive and multi-task benchmark for protein sequence understanding

    Minghao Xu, Zuobai Zhang, Jiarui Lu, Zhaocheng Zhu, Yangtian Zhang, Chang Ma, Runcheng Liu, and Jian Tang. PEER: A comprehensive and multi-task benchmark for protein sequence understanding. InThirty-sixth Conference on Neural Information Processing Systems Datasets and Benchma...

    2022

  42. [50]

    PETA: evaluating the impact of protein transfer learning with sub-word tokenization on downstream applications.Journal of Cheminformatics, 16(1):92, 2024

    Yang Tan, Mingchen Li, Ziyi Zhou, Pan Tan, Huiqun Yu, Guisheng Fan, and Liang Hong. PETA: evaluating the impact of protein transfer learning with sub-word tokenization on downstream applications.Journal of Cheminformatics, 16(1):92, 2024

    2024

  43. [51]

    ProteinGLUE multi-task benchmark suite for self-supervised protein modeling

    Henriette Capel, Robin Weiler, Maurits Dijkstra, Reinier Vleugels, Peter Bloem, and K Anton Feenstra. ProteinGLUE multi-task benchmark suite for self-supervised protein modeling. Scientific Reports, 12(1):16047, 2022

    2022

  44. [52]

    DeepLoc: prediction of protein subcellular localization using deep learning

    José Juan Almagro Armenteros, Casper Kaae Sønderby, Søren Kaae Sønderby, Henrik Nielsen, and Ole Winther. DeepLoc: prediction of protein subcellular localization using deep learning. Bioinformatics, 33(21):3387–3395, 2017

    2017

  45. [53]

    DeepSol: a deep learning framework for sequence-based protein solubility prediction

    Sameer Khurana, Reda Rawi, Khalid Kunji, Gwo-Yu Chuang, Halima Bensmail, and Raghven- dra Mall. DeepSol: a deep learning framework for sequence-based protein solubility prediction. Bioinformatics, 34(15):2605–2613, 03 2018

    2018

  46. [54]

    FLIP: Benchmark tasks in fitness landscape inference for proteins

    Christian Dallago, Jody Mou, Kadina E Johnston, Bruce Wittmann, Nick Bhattacharya, Samuel Goldman, Ali Madani, and Kevin K Yang. FLIP: Benchmark tasks in fitness landscape inference for proteins. InThirty-fifth Conference on Neural Information Processing Systems Datasets and B...

    2021

  47. [55]

    ProteinShake: Building datasets and benchmarks for deep learning on protein structures

    Tim Kucera, Carlos Oliver, Dexiong Chen, and Karsten Borgwardt. ProteinShake: Building datasets and benchmarks for deep learning on protein structures. InThirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2023

    2023

  48. [56]

    Using support vector machine combined with auto covariance to predict protein–protein interactions from protein sequences

    Yanzhi Guo, Lezheng Yu, Zhining Wen, and Menglong Li. Using support vector machine combined with auto covariance to predict protein–protein interactions from protein sequences. Nucleic Acids Research, 36(9):3025–3030, 2008

    2008

  49. [57]

    Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning.Nature Methods, 17(2):184–192, 2020

    Pablo Gainza, Freyr Sverrisson, Frederico Monti, Emanuele Rodola, Davide Boscaini, Michael M Bronstein, and Bruno E Correia. Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning.Nature Methods, 17(2):184–192, 2020

    2020

  50. [58]

    DIPS-plus: The enhanced database of interacting protein structures for interface prediction.Scientific Data, 10(1):509, 2023

    Alex Morehead, Chen Chen, Ada Sedova, and Jianlin Cheng. DIPS-plus: The enhanced database of interacting protein structures for interface prediction.Scientific Data, 10(1):509, 2023

    2023

  51. [59]

    A comprehensive dataset of protein-protein interactions and ligand binding pockets for advancing drug discovery.Scientific Data, 11(1):402, 2024

    Alexandra Moine-Franel, Fabien Mareuil, Michael Nilges, Constantin Bogdan Ciambur, and Olivier Sperandio. A comprehensive dataset of protein-protein interactions and ligand binding pockets for advancing drug discovery.Scientific Data, 11(1):402, 2024

    2024

  52. [60]

    The PDBbind database: methodologies and updates.Journal of Medicinal Chemistry, 48(12):4111–4119, 2005

    Renxiao Wang, Xueliang Fang, Yipin Lu, Chao-Yie Yang, and Shaomeng Wang. The PDBbind database: methodologies and updates.Journal of Medicinal Chemistry, 48(12):4111–4119, 2005

    2005

  53. [61]

    Fragment

    Raphael JL Townshend, Martin Vögele, Patricia Suriana, Alexander Derry, Alexander Powers, Yianni Laloudakis, Sidhika Balachandar, Bowen Jing, Brandon Anderson, Stephan Eismann, et al. Atom3d: Tasks on molecules in three dimensions.arXiv:2012.04035, 2020. 13 A Limitations and B...

    2012 arXiv

  54. [62]

    Task” indicates evaluation scope: “All

    are pretrained protein language models using masked language modeling on large sequence corpora. ANKH[ 29] and PROTT5 [ 28], in contrast, adopt encoder–decoder architectures, enabling bidirectional contextualization and autoregressive decoding. While TM-VEC[ 32] and PROSTT5 [ ...

Pith tools

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