Conference Day Two
Streamlining Drug Development with Advanced Bioinformatics
8:30 am Welcome Coffee
8:50 am Chair’s Opening Remarks
9:00 am Optimizing your Biologic’s Design for Downstream Developability
Synopsis
- Contextualizing antibody developability parameters through use cases, with focus on antibody hydrophobicity and polyspecificity
- Leveraging deep learning to learn patterns in antibody sequences and feed into developability prediction
- Combining NGS data and deep learning to inform synthetic library designs yielding antibodies with better developability
9:30 am Panel Discussion: Using Deep & Machine Learning Methods to Develop & Optimize Therapeutic Anti-bodies
Synopsis
- Revealing the pillars for ML-based optimization of therapeutic antibodies
- Outlining recent progress in the application of ML methods for antibody optimization
- Comparing organizational strategies for developing optimal therapeutic antibodies
10:15 am
Morning Networking & Coffee
10:45 am Optimising Antigen Specificity Using Computational Structural Modelling
Synopsis
- Evaluating current antibody and antigen structure models and their contribution to structure-based design
- Exploring challenges when applying structural analytics via a practical case study
- How far are we from designing antibodies fully in silico?
11:15 am Driving Antibody Optimization Using a Mix of Sequence, Structure, Computation, and Screening
Synopsis
- Addressing multi-goal lead optimization: Developability, immunogenicity, affinity, specificity
- When do you design from sequence, model structure, search computationally, and/or screen libraries?
- Discussing ATRC-101 and ATRC-301 clinical candidate optimization case studies
11:45 am Roundtable: Addressing In-Silico Prediction of Immunogenicity to Circumvent Alterations in Drug Pharmacokinetics, Pharmacodynamics, & Efficacy
Synopsis
- Revealing the latest developments for in-silico immunogenicity prediction
- Outlining examples and the different ways biologics can be immunogenic
- In what cases are computational tools meaningful? Where do they fail?
12:30 pm
Networking Lunch Break
1:30 pm Acclerating AI and Data-driven Design of Biologics to Untap Novel Biotherapeutics
Synopsis
- Examining best-practices for data-driven design of biologics by showcasing the optimization of a bispecific antibody
- How can we best leverage AI to design novel biotherapeutics?
- What the future hold for biologics in the age of AI?
CDD 2.0 – De-risking Your Novel Biologics Pipeline Progression
2:00 pm Applying Natural Language Processing (NLP) to Help Model Informed Biologic Development
Synopsis
- Review applications of NLP algorithms in different stages of drug development lifecycle through NLP use cases
- Overcoming challenges to NLP employment, including reproducibility and explainability, to improve NLP acceptability by researchers and regulators
- Minimising bias in NLP models to motivate wider adoption of NLP technologies
2:30 pm Panel Discussion: Facilitating End-To-End Integration of AI from Early Biologic Discovery to Late Development
Synopsis
- Examining common challenges for end-to-end integration of AI within your organisations
- How can we foster collaboration between different departments, including research, development, and CMC,
and establish AI as a core organisational discipline? - What are the best-practices for incorporating integrated platforms for predictive models for from discovery to development?
3:15 pm
Afternoon Refreshments
3:45 pm Unpacking a Workflow for Automating Protractor Placement on Peptides for Enhanced Half Life
Synopsis
- Employing crude heuristics for protractor placement on peptide scaffolds
- Introducing a workflow that considers spurious interactions with protractors that may degrade the peptide or the interface
- Can peptide-protractor interaction hotspots be leveraged for formulation design?
4:15 pm Leveraging Protein Language Modeling for Antibody Developability Prediction
Synopsis
- Exploring the pretrained protein language models for antibody developability prediction
- Investigating the possibility of transfer learning from pretrained protein language models to alleviate the scarcity of antibody developability data
- Proposing a rapid and high throughput framework for antibody developability prediction using sequence information alone