[scrollsequence id="135825"]
Designing antibodies from scratch can unlock novel biology and lead to breakthrough biologics in a fraction of the time of traditional drug discovery.
But it's only possible with a wet lab and AI models working together in an iterative loop.
Data is the bottleneck of AI drug creation
Large language models have access to the entire internet to train on, but expansive data sets don't exist yet for AI biologics discovery.
Since 2020, Absci has been amassing an exponentially larger data moat
We use this to train our AI models to uncover novel biology and de novo design breakthrough biologics.
[scrollsequence id="135825"]
Data to train
We've built a 77,000+ sq ft wet lab to generate high-quality biological training data with proprietary data generation technologies.
Proprietary Datasets
SoluPro® is our synthetic biology technology that generates billions of cells expressing proteins of interest.
Our ACE Assay then screens millions of antibody sequence variants with billions of parameters at >4,000x throughput compared to traditional assays.
Public Datasets
Combined with publicly available biological data, we have built enormous sets of specialized training data for AI model development.
AI to create
Built on these massive data sets, our generative AI models create next-gen biologics using de novo design and AI optimization.
De novo antibody design
Zeroing in on epitopes of interest on a target antigen, our generative AI engine creates millions of novel antibody designs in silico.
Designing antibodies to bind to specific epitopes allows us to test new hypotheses and unlock differentiated biology that leads to better drug designs.
AI lead optimization
From de novo designed leads, we AI optimize for target affinity, potency, safety, manufacturability, half-life, and engineer-in other properties for a potential best-in-class profile.
Wet Lab to validate
With massive throughput functional validation from our wet lab, we're able to assess binding affinity and target specificity for millions of unique AI-generated designs a week.
Data Flywheel
The data we generate from wet lab validation propels an experimental cycle that takes us from data to train, AI to create, and wet lab to validate new therapeutic designs in as little as six weeks.
Data Engineering
Our datasets have been growing since 2020 and the lines connecting the learning loop are just as pivotal. Our AI is supported by a dedicated data engineering team that ensures we're scaling our data and models to improve exponentially.
Target to promising leads in just over a year
Our Integrated Drug Creation platform can unlock differentiated biology and advance AI-designed and optimized development candidates in as few as 14 months - and potentially faster as our data and AI models continue to progress.
Target to promising leads in just over a year
Our Integrated Drug Creation platform can unlock differentiated biology and advance AI-designed and optimized development candidates in as few as 14 months - and potentially faster as our data and AI models continue to progress.
case study
AI de novo designed and optimized TL1A antibody
Absci applied our Integrated Drug Creation platform to design a differentiated anti-TL1A antibody for Inflammatory Bowel Disease (IBD) with potential best-in-class properties.
ai target discovery
Discovering novel targets with Reverse Immunology
Our Reverse Immunology approach identifies novel therapeutic targets from patient samples of super immune responders. The target, paired with a fully human monoclonal antibody, serves as a starting point for AI optimization and development.
