Computational Antibody Papers

Filter by tags
structure prediction
Filter by published year
All
TitleKey points

  • 2025-06-05

    PSBench: a large-scale benchmark for estimating the accuracy of protein complex structural models:

    • structure prediction
    • PSBench is a large benchmark dataset (>1M models) for training and evaluating model accuracy estimation (EMA) methods for protein complex structures, using data from CASP15 & CASP16.
    • Models were generated by AlphaFold2-Multimer and AlphaFold3 under blind prediction conditions and annotated with 10 detailed global, local, and interface quality scores.
    • The dataset enables development of advanced EMA methods (e.g. GATE), which showed top performance in blind CASP16 assessments.

  • 2025-03-31

    AI-Augmented Physics-Based Docking for Antibody-Antigen Complex Prediction

    • epitope prediction
    • docking
    • structure prediction
    • Benchmarking of the structure prediction/docking and co-folding methods for antibody design
    • Authors measure the impact of antibody-antigen model quality on the success rate of epitope prediction and antibody design.
    • For epitope prediction and antibody design they use a proxy measure of DockQ score - they call success when DockQ is better than 0.23, for antibody design they use a stricter threshold of 0.49.
    • Using these measures, AlphaFold3 comes out on top, and it would be successful roughly ~47% times.
    • THey introduce an approach where ProPOSE and ZDOCK decoys are refined using AlphaFold. With this combined protocol they reach success rates of 35% for epitope mapping and 30% for antibody design.

  • 2025-03-11

    Fast and accurate antibody sequence design via structure retrieval

    • databases
    • generative methods
    • structure prediction
    • Inverse folding and thus antibody design via database search.
    • Authors train a vector retrieval database on SAbDab. In this way for a single sequence one can figure out where it falls structurally.
    • They benchmark against state of the art inverse folding tools such as AbMPNN, AntiFold, ProteinMPNN and ESM-IF - their tools comes on top in terms of sequence retrieval.
    • The database search is orders of magnitude faster than the state of the art inverse folding tools.
    • They compare IgSeek versus FoldSeek - their tool gets a higher accuracy in sequence retrieval, for most CDRs, but CDR-H3. Therefore FoldSeek seems like a very good choice alongside IgSeek for such a database-driven inverse folding protocol.

  • 2024-10-24

    Chai-1: Decoding the molecular interactions of life

    • structure prediction
    • Foundational model following in the footsteps of AlphaFold3 attempting prediction of molecular interactions.
    • Model architecture is closely modeled on this of AF3 - however it was not benchmarked against it (nor ESM3) because of use restrictions.
    • It takes 30 days on 128 A100s to train the model. Back of the envelope Google Colab pro+ pricing of 128 (which is NOT distributed) puts it at ca. 120k USD :)
    • Addressing antibodies, they introduce constraints (e.g. known epitope residue) to help the model out. Adding even a single residue makes a big difference with respect to baseline which is tantamount to global docking.
    • It only takes one residue as constraints to improve ab-ag complex prediction. Success rate for ‘local’ mode, with just one residue is about 50% in getting it with DockQ score >.21, 30% >.49 and less than 10% for high quality hits.
    • Success rate for ‘global’ mode, so without constraints is about 35% in getting it with DockQ score >.21, 20% >.49 and less than 5% for high quality hits.
    • So altogether if I want to ‘hit the epitope’, the model has ca. 30% success rate.
    • If I want a high quality ab-ag complex structure, unfortunately it seems that constraints do not help much currently.

  • 2024-10-08

    Accurate antibody loop structure prediction enables zero-shot design of target-specific antibodies

    • generative methods
    • structure prediction
    • Authors describe how using a structure predictor one can re-design the binding site, to maintain binding.
    • They use a proprietary GaluxDesign method, the method achieves 1.4 Å Ca RMSD in predicting CDR-H3 loop structures, leveraging a unique scoring metric (G-pass rate) that assesses both confidence and structural consistency for antibody design.
    • The method outperforms AlphaFold 2.3, ABlooper, and ImmuneBuilder in predicting CDR-H3 loop structures, with significantly lower RMSD values (1.4 Å compared to 2.4-3.7 Å), particularly on a more challenging, time-separated dataset.
    • The binding propensity to HER2 was evaluated using a large mutant library and calculated via the G-pass rate, outperforming AlphaFold's PAE-based scoring. The model showed strong discrimination with an AUROC of 0.758, compared to 0.529 for AlphaFold. The novel loop is scored using their metric (G-pass rate) in complex with Her2.
    • Novel antibody sequences were designed by predicting six CDR loops in antibody-protein complexes, using GaluxDesign models. These designs were experimentally tested, achieving high success rates, including a 13.2% success rate for HER2 antibody designs using yeast display methods.

  • 2024-09-04

    Protein loop structure prediction by community-based deep learning and its application to antibody CDR H3 loop modeling

    • structure prediction
    • Novel CDR-H3 structure prediction method, ComMat based on ensemble sampling.
    • Rather than generating a single structure, the method generates several solutions that are then all informing the next iteration.
    • The method was integrated into the structure module of AlphaFold2.
    • Crucially, with the introduction of the second prediction into the ‘community’, the predictions become better. However these quickly plateau, showing the limits of the approach.
    • The method does not produce better results than ABodyBuilder2 and EquiFold.

  • 2024-08-21

    ABodyBuilder3: Improved and scalable antibody structure predictions

    • structure prediction
    • language models
    • Updated version of the popular ABodyBuilder2 program to model antibodies, that is more efficient.
    • Unlike ABB2 that did not employ language model embeddings, ABB3 does (ProtT5 to be precise).
    • Instead of using several models and aggregating the results for error prediction, they train pLDDT within the model.
    • ABB3 achieves CDR-H3 RMSD in the ballpark of 2.4Å whereas the previous version in the region of 2.5Å.

  • 2024-07-30

    Fast and accurate modeling and design of antibody-antigen complex using tFold

    • binding prediction
    • structure prediction
    • docking
    • Update on tfold-AB including modeling of the complex with the antigen, with applications to virtual screening.
    • They mostly use SABDAB/Covabdab as reference datasets.
    • The modeling happens by generating antibody & antigen features supplemented by a large language model, followed by flexible docking.
    • For antigen feature generation, they use AF2
    • They are training on several tasks simultaneously, ab structure prediction, complex prediction etc. making it a multi-task training.
    • On docking their method achieves DockQ 0.217 vs AlphaFold-Multimer DockQ score of 0.158 - that is global docking.
    • Whe local docking information is given, constraining paratope/epitope sites, their algorithm achieves DockQ of 0.416.
    • They demonstrate that filtering antibodies by their predicted modeling confidence score gives moderate enrichment against PD1 and Sars-cov-2 antigens, showing promise for virtual screening.

  • 2024-06-25

    Improved computational epitope profiling using structural models identifies a broader diversity of antibodies that bind the same epitope

    • structure prediction
    • ngs
    • An update on SPACE1, employing ABodyBuilder2. Better coverage of the structural method.
    • They used binders against coronavirus, ebola, lysozyme among others.
    • Structures are modeled using ABodyBuilder2. Structures are sorted by CDR lengths, frameworks aligned by Cas and RMSD calculated for CDR loops. Finally a clustering algorithm is used.
    • The clustering algorithms benchmarked were DBSCAN, OPTICS-xi, OPTICS-DBSCAN, K-means, Butina clustering, greedy clustering.
    • Two more variants were developed, SPACE2-HC, for heavy chains only as well as SPACE2-Paratope, for paratyping.
    • Two accuracy metrics were used, the fraction of epitope-consistent clusters (number of epitope-consistent multiple-occupancy clusters / number of multiple-occupancy clusters) and the fraction of clustered antibodies in epitope-consistent clusters (number of antibodies in epitope-consistent multiple-occupancy clusters / number of antibodies in multiple-occupancy clusters)
    • Two coverage metrics were used, the number of multiple-occupancy clusters and the number of antibodies in multiple-occupancy clusters were used. In order to examine accuracy and coverage with one measure they calculated the number of antibodies in consistent multiple-occupancy clusters
    • They selected agglomerative clustering as best, though it is not better than Optics-XI, but it was providing larger clusters.
    • Space2 using all loops was better than SPACE2-HC or SPACE2-paratope
    • Space2 improves the coverage over SPACE1, thanks to the ABodyBuilder2 protocol.
    • Space2 increases coverage with respect to just clonotyping, but clonotyping remains much more accurate.

  • 2024-03-13

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

    • structure prediction
    • language models
    • Introducing ESM-2 and ESMFold. Scaling transformer model parameter size to 15B allows for more precise predictions of structures.
    • They make available an atlas of 617 million predicted structures
    • Learning objective is MLM, masking 15% of protein input.
    • Perplexity ranges from 1 for a perfect model to 20 for a model that makes predictions at random. Intuitively, perplexity describes the number of amino acids the model is uncertain between when it makes a prediction
    • After 270k training steps the 8M parameter model has a perplexity of 10.45, and the 15B model reaches a perplexity of 6.37.
    • The 15B model achieves best perplexity and structural modeling accuracy.
    • For some structures, accuracy of structure prediction jumps from 7.7A at 8m parameters to 7.0A at 35m parameters and to 3.2A at 15m parameters. The 3b model brings it down to 2.8 and 15B model to 2.6. For other structures, good prediction is only achieved at 15B
    • Their structure predictor closely follows AlphaFold2, but instead of evoformer, they use the representation from the ESM-2.