SC8: Engineering Better Antibody Therapeutics Using Computational Approaches
SUNDAY, MAY 3 | 2:30 – 5:30 PM
ABOUT THIS COURSE:
This course will cover the Oxford Protein Informatics Group (OPIG) tools for therapeutic antibody research, including our structural antibody databases (SAbDab, Thera-SAbDab), our antibody structure prediction suite (SAbPred), and our antibody next-generation sequencing database (OAS). We will show how the vast numbers of antibody sequences in databases like OAS can be structurally annotated (SAAB+) or modeled (ABodyBuilder). Suggested applications will cover repertoire mapping, developability issue prediction (TAP), and screening library design. No prior programming experience is required; we encourage attendees to bring their laptops to explore our web applications during the workshop.
WHAT YOU WILL LEARN:
- This course will demonstrate how computational approaches can aid drug discovery – in particular, how antibody sequences can be transformed into three-dimensional structures, and the associated benefits.
- The course will contain the essential theory needed to understand how our tools work, as well as usage demonstrations.
- A booklet describing all our programs (with example case studies) will be provided in advance of the course and can serve as a reference for future use.
- We will share how to obtain our tools in a browser-based or command-line format through our “SAbBox” Virtual Box service.
COURSE AGENDA:
2:30 Antibody Sequences and Numbering
Aleksandr Kovaltsuk, MSc (Glas.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Overview: Brief recap of antibody function and genetic origin. Variable domain sequences and numbering schemes. Framework and complementarity determining region (CDR) definitions. Demonstration of ANARCI for numbering variable domains. Applications.
3:00 Antibody Structures and Homology Modeling
Matthew I. J. Raybould, MChem (Oxon.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Overview: Why is structure important? Antibody variable domain structure. The Protein DataBank and PDB file format. Demonstration of SAbDab (structural antibody database) and Thera-SAbDab (therapeutic structural antibody database).
The need for antibody modeling. Heavy/light chain orientation prediction using ABangle. Canonical forms and their prediction from sequence (SCALOP). Using FREAD for loop homology modeling and grafting, and Sphinx for hybrid homology/ab initio loop modeling. Using PEARS for antibody-specific side chain prediction. Finally, a demonstration of ABodyBuilder, the complete pipeline for full variable domain modeling. Statistical confidence metric.
3:45 Refreshment Break
4:05 Structural Annotation of Antibody Next-Generation Sequencing datasets for “humanness” assessment and immunodiagnostics
Aleksandr Kovaltsuk, MSc (Glas.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Overview: Demonstration of the Observed Antibody Space (OAS) database of > 1Bn variable domain antibody sequences. Includes the ability to search by individual, disease-state, species, isotype, etc. Followed by a description of the SAAB+ tool for structurally annotating very large singlechain datasets, with example applications.
4:30 Short Break
4:35 Structural Modeling of Antibody Next-Generation Sequencing datasets for developability assessment and screening library design
Matthew I. J. Raybould, MChem (Oxon.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Overview: A workflow that creates full Fv model structural summaries (“Antibody Model Libraries”) from antibody NGS datasets. Applications for therapeutic developability assessment (the Therapeutic Antibody Profiler, with demonstration), and general or target-focused screening library design.
5:10 Opportunity for Questions/Consolidation
5:25 Course wrap-up
Including information on the SAbBox Virtual Box, which contains all the software/databases described in the workshop.
5:30 Close of Course
INSTRUCTOR BIOGRAPHIES:
Aleksandr Kovaltsuk, MSc (Glas.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Aleksandr Kovaltsuk received his MSc degree in Pharmacology from the University of Glasgow in 2016, conducting his research year in the Antibody Discovery and Protein Engineering department at MedImmune (AstraZeneca) in Cambridge, UK. He then joined the Oxford Protein Informatics Group at the University of Oxford to study for a DPhil under the supervision of Professor Charlotte Deane in collaboration with UCB Pharma. In 2017 he was awarded a Royal Commission Industrial Fellowship. He authored the Observed Antibody Space (OAS) antibody NextGeneration Sequencing (NGS) database, and the ABOSS and SAAB+ tools for errorfiltering and structurally annotating antibody NGS datasets respectively. He has written a review on the benefits of structural antibody NGS analysis. His current project involves developing software for immunodiagnostics and humanisation.
Matthew I. J. Raybould, MChem (Oxon.), DPhil Student, Oxford Protein Informatics Group, Department of Statistics, University of Oxford
Matthew Raybould received his MChem degree in Chemistry from the University of Oxford in 2016. He is currently studying for a DPhil, working in the Oxford Protein Informatics Group at the University of Oxford under the supervision of Professor Charlotte Deane. He collaborates with AstraZeneca, GSK, Roche, and UCB. He wrote the Therapeutic Structural Antibody Database (Thera-SAbDab) and the Therapeutic Antibody Profiler (TAP) for structure-based developability assessment. He has written reviews on harnessing antibody NGS data in therapeutic discovery. His research on antibodies currently focuses on in silico high throughput screening and subsequent rational optimisation.