Conference Day One
8:00 am Registration & Welcome Coffee
8:45 am Chair’s Opening Remarks
Bridging the Gap between Data & Model in Computational Biology
9:00 am Harnessing Artificial Intelligence to Make Antibodies Smarter
Synopsis
- Delivering the story of the AI-designed antibody, AU-007, from idea to in-vivo to patient
- How can we give antibodies dynamic capabilities in humans while maintaining the developability profile?
- What data do we need to design smarter biotherapeutics?
9:30 am Accelerating Drug Discovery and Development with a Virtual Twin Approach
Synopsis
- Reviewing the importance of Computational Biologics Design and Development (CBDD) approaches to address the issue of multi-objective screening & multi-dimensional optimization
- Discussing how engineering of biologics has led to an increase in the variety of formats. Optimization of biologics is becoming more complex
- Highlighting workflows and the thought process to tackle these challenges posed by the complexity of the various antibody-based biologics
10:00 am Panel Discussion: Debating Best-Practices for Data Flow, Accumulation, & Utilisation to Provide Novel Insight into Biological Processes & Biologic Design
Synopsis
- Reviewing approaches for validation, storage & protection of data under the FAIR principles
- Navigating large biological data sets based on data warehouses for data integration & management provides novel insight into biological processes
- How do we create a database that is representative of a wide variety of biological information?
10:45 am
Speed Networking & Morning Coffee
11:15 am Session Reserved for CDD Vault
Synopsis
11:45 am From Data to Predictions: AI-Based Virtual Screening for Multi- Specific Protein Therapeutics
Synopsis
- Examining the setup of our industry-leading high-throughput engineering platform for multi-specific protein therapeutics
- Applying AI- and structure-based in-silico design workflows to extract knowledge from huge engineering data sets and to guide the discovery &
optimization of novel multi-specific biologics - Exemplifying an AI-based virtual screening approach within one of Sanofi’s multi-specific biologics optimization campaigns
12:15 pm Data-Driven Target Discovery: Leveraging Real World Human Data to Understand Disease Mechanism
Synopsis
- How can we use real world data to understand disease mechanisms?
- How can we leverage data to support the development of novel therapeutics?
- Assessing in-vivo & in-vitro preclinical models in downstream screening and validation
- Highlighting future avenues for MoA elucidation, whereby data is translated into meaningful hypothesis for novel biologics candidates through clinic
12:45 pm Enabling Robust Prediction of Protein-Protein Binding Affinities Using FEP+
Synopsis
- Physics-based free energy perturbation (FEP) calculations provide accurate energetics while allowing conformational flexibility by using explicit solvent molecular dynamics (MD) simulations with our state-of-the-art OPLS4 force field
- Enhanced sampling protocols for the mutating residue and nearby waters; intelligent handling of Proline and charged amino acids; and automated parameterization of noncanonical amino acids lead to improved results
- Both FEP+ and our new constant-pH molecular dynamics (CpHMD) implementation can account for protonation and tautomeric state changes, both upon binding/folding and at different pH values
1:15 pm
Networking Lunch Break
Accelerating Novel Biotherapeutics Discovery with Computational Biology
2:00 pm Developing a Data Management Workflow to Accelerate Candidate Selection & Reduce Risks in Development
Synopsis
- Introducing Merck’s novel data management system, now usable for ML/AI with millions of entries
- Integrating the Hight Throughput analytical characterization data management system “Spotfire Dashboard” to drive better decision-making when screening through a large dataset
- Examining best practices for data selection and curation to achieve maximum impact during the early stages of antibody molecule generation
2:30 pm Applying AI to Unravel the Massive Amounts of Highly Organised Protein Data Stored in 3DM’s Protein Superfamilies Databases
Synopsis
- What is 3DM and what makes is it unique to other methods?
- Why is Helix, the AI platform build on 3DM, so much better in predicting the effects of Human SNPs than all other available tools?
- How can you as customer access the predictions including all underlying data and reports that explain why a mutation is predicted to be pathogenic or not?
- How do we apply the same strategy to speed up protein engineering projects, including optimising or humanizing antibodies?
3:00 pm Exploring a Theoretical Framework to Guide De Novo Biologic Design
Synopsis
- Task 1: Given the antigen sequence, can you set the sequence motif of the antibody to make a reliable prediction on whether a particular antibody would bind to this antigen?
- Task 2: Given an antibody sequence, can you identify the most likely epitope on a particular antigen if you already know that this antibody binds to this antigen?
- Task 3: Given a known antibody sequence that binds to a particular epitope on an antigen, how many different antibody sequences we identify that will bind to the same epitope?
- Task 4: Given an arbitrary antigen, can we identify one or a few epitopes that can help design an antibody sequence binding to the designated epitope?
3:30 pm Predicting Antibody Promiscuity Based on Physico-Chemical Bio- Properties Through Integration Of Internal Peripheral Blood Cell (PBMC) Binding Data
Synopsis
- Capturing existing peripheral blood mononuclear cell (PBMC) binding data sets to build models
- Applying machine learning methods to predict antibody promiscuity (PBMC binding)
- Identifying antibody physicochemical properties to predict and understand promiscuity for prospective antibodies
4:00 pm
Afternoon Refreshments
4:30 pm Addressing the Design & Engineering of Multi-specific, Multifunctional Antibodies to Advance the Constructs in-Vivo and into the Clinic
Synopsis
- Discussing advances in the development of multifunctional antibodies with favourable stability, specificity, and pharmacokinetic properties
- Exploring Emerging computational methods for enhancing the next generation of multi-specific antibodies
- Leveraging current know-how by translating learnings from IgG1 design to multispecifics
5:00 pm Unravelling Computational Prediction of Protein–Protein Interactions
Synopsis
- Reviewing the current tools and gaps within protein–protein complex prediction
- How do you rank order the different poses that result from docking?
- How is machine learning impacting the field of protein-protein interactions?
- Examining applications of antigen-antibody docking for epitope identification/ epitope binning