Computational Antibody Papers

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

  • 2024-08-21

    Training data composition determines machine learning generalization and biological rule discovery

    • binding prediction
    • Authors investigate the role of the negative/positive binding data composition on predictive power of ab-ag interaction prediction.
    • They employ Absolut! Framework to generate synthetic binding data.
    • For binding prediction they use a 10 hidden neuron fully connected network.
    • CDR-H3 is used as a proxy for binding for computational expediency.
    • Altogether, if data in positive and negative are quite similar, there is better chance to out of distribution generalization.

  • 2024-08-15

    Active learning for affinity prediction of antibodies

    • binding prediction
    • experimental techniques
    • An active learning framework is proposed to efficiently identify antibody mutations that enhance binding affinity, minimizing wet-lab experiments.
    • From the paper: “Active learning is a framework from experimental design that focuses on making informed decisions about which experiments to perform next”. That’s important these days to effectively generate prediction first data to maximize the effectiveness of models.
    • Bayesian optimization is used with relative binding free energy (RBFE) methods to iteratively propose and evaluate new antibody sequences, improving binding affinity predictions.
    • Various encoding schemes, including one-hot, bag of amino acids, BLOSUM, and AbLang2, are tested. The best performing methods are identified through validation with pre-computed data.
    • The study uses two RBFE methods: NQFEP for accurate but costly simulations, and Schrödinger Res Scan for faster but less precise results. The active learning loop consistently finds better binding sequences using these methods.
    • AbLang2 encoding with the Tanimoto kernel consistently outperformed other methods in the validation phase, indicating its effectiveness in predicting improved binding affinities.
    • NQFEP method provides more accurate but computationally expensive simulations, while the Schrödinger Res Scan offers quicker but less precise results. Despite the higher computational cost, the NQFEP method might be preferable when high accuracy is crucial.

  • 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.

  • 2024-07-23

    Baselining the Buzz Trastuzumab-HER2 Affinity, and Beyond

    • databases
    • protein design
    • generative methods
    • binding prediction
    • Novel dataset of 0.5m anti-trastuzumab antibodies together with benchmarking of affinity classification methods.
    • They generated a dataset of ~500,000 anti-Her2 trastuzumab variants by modifying the CDR-H3. The binding affinity is divided into high/medium/low with reasonably even split (178,160, 196,392, and 171,732 respectively).
    • They split their dataset into positives and negatives by putting medium and low binders into the negative set.
    • They contrast their dataset with this from Mason et al. (~39k variants vs theirs 0.5m), to show that in a small (relatively) number of cases, binders in one set can be labeled as negative in another.
    • They test that the predictor developed by Mason et al. for binding/non binding classification works as intended on the novel 0.5m dataset. Likewise their model trained on Mason data and tested on their ds (and vice versa) has predictive power. It does but to a much lesser extent than training on data from the same experiment.
    • As methods to benchmark they used FLAML https://arxiv.org/abs/1911.04706, CNN and EGNN.
    • CNN and FLAML are top performers, but CNN performs well on the small data (signal starting with ~170 sequences).
    • Performance drops radically when train/val splits are done with respect to the clonotype.
    • THey tested AbLang, ProteinMPNN, ESM and Blosum on their ability to generate binding trastuzumab variants. As control they also randomly generated sequences. They observed the following percentages of sequences exhibiting CNN-HER2-max binding probabilities greater than 90%: 13% for Random, 26% for BLOSUM, 27% for AbLang (when masking all ten residues simultaneously), 29% for AbLang (when masking one residue at a time), 19% for ProteinMPNN, and 30% for ESM (when masking one residue at a time), respectively.

  • 2024-07-23

    Atomically accurate de novo design of single-domain antibodies

    • binding prediction
    • protein design
    • Novel method to design antibodies, validated on VHHs.
    • The protocol includes an antibody-fine tuned RFdiffusion to create the coordinates for the backbone followed by creating the sequence using ProteinMPNN (not antibody-fine-tuned). Filtering by RoseTTafold2 can be used.
    • Fine-tuning RFDiffusion includes giving the coordinates of the antigen, framework of the antibody and information on the epitope. Only the CDR coordinates and orientation of the framework/CDRs are designed.
    • They fine-tuned RoseTTafold2 with the aim to filter out the RFdiffusion designs.
    • The fine-tuned RoseTTafold2 can accurately distinguish correct and incorrect ab-ag pairs, but when hot-spot information is provided (aka local docking).
    • The best designs range from micromolar and high nanomolar and roughly 1 in 100 designs works as intended : https://www.nature.com/articles/d41586-024-00846-7

  • 2024-07-16

    PeSTo: parameter-free geometric deep learning for accurate prediction of protein binding interfaces

    • binding prediction
    • epitope prediction
    • Geometric transformer that receives a single structure at input and annotates residues with likelihood that they are part of a binding site.
    • The geometric transformer only uses the atom names, no mass, charge etc.
    • Similar to convolution their geometric attention mechanism focuses on 8 nearest neighbors (3.2A) and then increases up to 64 nn (8.2A).
    • They use ~300,000 chains from the PDB for training (!). This is because they have extracted all the bio assemblies at 30% sequence identity.
    • They defined the most common atom names for all molecule types which gave them 79 elements. Interactions between these elements can be given as a 79x79 matrix.
    • Interaction cutoff is taken as 5A.
    • Pesto outperforms Scannet by a large margin, 0.93 vs 0.87 ROC AUC.
    • In some cases, processing MD trajectories of unbound proteins with PeSTo identifies certain interfaces better than when PeSTo is run on the starting static structure.

  • 2024-07-09

    Investigating the Volume and Diversity of Data Needed for Generalizable Antibody-Antigen ∆∆G Prediction

    • binding prediction
    • They created a siamese EGNN, one given WT the other one mutant, with their difference being the ddG prediction.
    • They used the AB-Bind dataset which consisted of 645 mutants from 29 complexes.
    • They created a set of non-redundant antibody-antigen binders with 1475 complexes. They imposed 70% clustering on antigens.
    • They mutated one complex per cluster and ran foldx resulting in 942,723 ddG foldX data points.
    • On ABBind dataset they achieve a pearson correlation of 0.8 - however when they impose stringent CDR cutoffs the correlation drops dramatically, indicating overtraining.
    • When they run the training on the synthetic dataset, it stops being sensitive to overtraining.

  • 2024-07-09

    Improved prediction of protein-protein interactions using AlphaFold2

    • binding prediction
    • docking
    • Using AF2 they developed a pipeline to fold and dock proteins simultaneously. The pipeline shows good performance in distinguishing interacting and non-interacting proteins.
    • Acceptable models are those with DockQ > 0.23. Success rate is defined as percentage of acceptable poses.
    • The best version of their model achieves a 39.4 success rate.
    • AlphaFold2 outperforms other docking methods.
    • Using the number of Cb in contact (within 8A) or plDDT of the interface results in ROC AUC in the region of .9 distinguishing interacting and non-interacting proteins.
    • As input they insert a chain break of 200 residues to model the interaction.
    • They note that it is very important to create the right MSAs for AF2.
    • As negative cases for interactions (non-interacting proteins) they employ data from Negatome.

  • 2024-07-02

    Surface ID: a geometry-aware system for protein molecular surface comparison

    • binding prediction
    • They draw from the Masif method in that they define a triangular mesh. Each vertex is encoded with physicochemical information and then each patch of a defined radius is encoded numerically.
    • They teach overlapping patches to have similar embeddings as they are assumed to have overlapping functions as well.
    • They employed contrastive learning, annotating patches as positive if they were within 1.5A from centered vertices and negative if they were centered on vertex 5A away.
    • Their learned similarity distances cluster by curvature, hydropathy and charge.
    • They compared Surface ID to structure based similarity measurement approaches with SurfaceID performing slightly better.
    • They clustered the antibody-epitope patches simultaneously. It clustered the binding modes between HIV-1 GP120, two for influenza HA and one for SARS-CoV-2 RBD. Anti-ha clustered had same epitope but different paratopes showing that the algorithm distinguishes on that level
    • They proposed a design scheme for antibodies. Look for similar epitopes by surface id and use the antibodies as putative binders to the query.

  • 2024-07-02

    AbAgIntPre: A deep learning method for predicting antibody-antigen interactions based on sequence information

    • binding prediction
    • Using a siamese network and Sabdab to predict antibody-antigen binding in a binary fashion.
    • They clustered the antigens at 0.9 sequence identity. They assumed that similar antibodies from the same antigen group bind in the same manner. This resulted in 3,892 antigen pairs.
    • They also created a dataset of covid specific antibodies with 9309 positive samples and 1710 negatives.
    • They used tha CKSAAP encoding, but compared against others such as one-hot, pssm or their-own trained word2vec.
    • They benchmark the different encodings and models to show that CKSAAP + CNN come out on top.
    • Their siamese CNN with CKSAAP achieve a staggering .85 PR AUC.