A5091100763- Monoclonal and Polyclonal Antibodies Research
- vaccines and immunoinformatics approaches
- Protein purification and stability
- Glycosylation and Glycoproteins Research
- CAR-T cell therapy research
- SARS-CoV-2 and COVID-19 Research
- Bacteriophages and microbial interactions
- Cell Image Analysis Techniques
- Genetics, Bioinformatics, and Biomedical Research
- Machine Learning in Bioinformatics
- Advanced Biosensing Techniques and Applications
- Computational Drug Discovery Methods
Bioscientifica (United Kingdom)
2023-2024
University of Oxford
2021-2023
BioScientia (Poland)
2023
Exscien (United States)
2023
Oxford Research Group
2021
Abstract In 2013, we released the Structural Antibody Database (SAbDab), a publicly available repository of experimentally determined antibody structures. interim, rapid increase in number structure depositions to Protein Data Bank, driven primarily by increased interest antibodies as biotherapeutics, has led us implement several improvements original database infrastructure. These include development SAbDab-nano, sub-database that tracks nanobodies (heavy chain-only antibodies) which have...
Antibodies are one of the most important classes pharmaceuticals, with over 80 approved molecules currently in use against a wide variety diseases. The drug discovery process for antibody therapeutic candidates however is time- and cost-intensive heavily reliant on vivo vitro high throughput screens. Here, we introduce framework structure-based deep learning antibodies (DLAB) which can virtually screen putative binding antigen targets interest. DLAB built to be able predict antibody-antigen...
Abstract Motivation Antibody-antigen complex modelling is an important step in computational workflows for therapeutic antibody design. While experimentally determined structures of both and the cognate antigen are often not available, recent advances machine learning-driven protein have enabled accurate prediction structures. Here, we analyse ability protein-protein docking tools to use learning generated input information-driven docking. Results In scenario, find that HADDOCK can generate...
Identifying the epitope of an antibody is a key step in understanding its function and potential as therapeutic. Sequence-based clonal clustering can identify antibodies with similar complementarity, however, from markedly different lineages but structures engage same epitope. We describe novel computational method for profiling based on structural modelling clustering. Using method, we demonstrate that sequence dissimilar functionally be found across Coronavirus Antibody Database, high...
Abstract Antibody-antigen binding affinity lies at the heart of therapeutic antibody development: efficacy is guided by specific and control affinity. Here we present Graphinity, an equivariant graph neural network architecture built directly from antibody-antigen structures that achieves state-of-the-art performance on experimental ∆∆G prediction. However, our model, like previous methods, appears to be overtraining few hundred data points available. To test if could overcome this problem,...
While conventional Transformers generally operate on sequence data, they can be used in conjunction with structure models, typically SE(3)-invariant or equivariant graph neural networks (GNNs), for 3D applications such as protein modelling. These hybrids involve either (1) preprocessing/tokenizing structural features input (2) taking Transformer embeddings and processing them within a representation. However, there is evidence that learn to process information their own, the AlphaFold3...
We introduce IgDiff, an antibody variable domain diffusion model based on a general protein backbone framework which was extended to handle multiple chains. Assessing the designability and novelty of structures generated with our model, we find that IgDiff produces highly designable antibodies can contain novel binding regions. The dihedral angles sampled show good agreement reference distribution. verify these designed experimentally all express high yield. Finally, compare state-of-the-art...
Antibody-antigen complex modelling is an important step in computational workflows for therapeutic antibody design. While experimentally determined structures of both and the cognate antigen are often not available, recent advances machine learning-driven protein have enabled accurate prediction structures. Here, we analyse ability protein-protein docking tools to use learning generated input information-driven docking. We find that HADDOCK can generate models antibodyantigen complexes using...
Abstract Developing therapeutic antibodies is a challenging endeavour, often requiring large-scale screening to produce initial binders, that still require optimisation for developability. We present computational pipeline the discovery and design of antibody candidates, which incorporates physics- AI-based methods generation, assessment, validation developable candidate against diverse epitopes, via efficient few-shot experimental screens. demonstrate these orthogonal can lead promising...
The field of protein-ligand pose prediction has seen significant advances in recent years, with machine learning-based methods now being commonly used lieu classical docking or even to predict all-atom complex structures. Most contemporary studies focus on the accuracy and physical plausibility ligand placement determine quality, often neglecting a direct assessment interactions observed protein. In this work, we demonstrate that ignoring interaction fingerprints can lead overestimation...
Abstract Antibodies are one of the most important classes pharmaceuticals, with over 80 approved molecules currently in use against a wide variety diseases. The drug discovery process for antibody therapeutic candidates however is time-and cost-intensive and heavily reliant on in-vivo in-vitro high throughput screens. Here, we introduce framework structure-based deep learning antibodies (DLAB) which can virtually screen putative binding antigen targets interest. DLAB built to be able predict...
We consider the problem of antibody sequence design given 3D structural information. Building on previous work, we propose a fine-tuned inverse folding model that is specifically optimised for structures and outperforms generic protein models recovery structure robustness when applied antibodies, with notable improvement hypervariable CDR-H3 loop. study canonical conformations complementarity-determining regions find improved encoding these loops into known clusters. Finally, applications...
Identifying the epitope of an antibody is a key step in understanding its function and potential as therapeutic. It well-established literature that sequence-based clonal clustering can identify antibodies with similar complementarity. However, there growing evidence from markedly different lineages but structures engage same near-identical binding modes. Here, we describe novel computational method for profiling based on structural modelling clustering, show how it sequence-dissimilar...
We introduce