EDEN™ Unprecedented precision for vaccine discovery of bacterial targets

Helping to combat the growing problem of antibiotic resistance, EDEN™ (Efficacy Discriminative Educated Network) is an innovative model that uses AI to identify antigens that can trigger a robust, protective immune response in patients. EDEN™ ranks the antigens based on their ability to elicit this response, and the most promising ones are then processed through (pre-)clinical pipelines before being formulated into vaccine products.

The core of our EDEN™ technology is a proprietary machine-learning ensemble of AI models used to interpret immunological-relevant information about bacterial antigens that incur protection in a vaccine setting. EDEN™ has been trained on our own curated data set derived to identify truly protective and non-protective antigens tested in clinical and pre-clinical settings. The input to the AI models is a feature transformation of the protein data set, in which several global and sequence-resolved properties are extracted. These structural and functional features have been selected for their relevance in protein chemistry, immunology, and protein structure and their ability to guide the network in discriminating protective versus non-protective antigens.

At a time when bacterial infections are becoming increasingly resistant to antibiotics, the development of new and effective vaccines is more important than ever. With EDEN™, the discovery of vaccines against bacterial diseases that were previously impossible to target is both highly scalable and fast, identifying antigens in as little as 48 hours.

Proteome identification: To identify novel, broadly protective vaccine antigens for a bacterial infection, EDEN utilizes proteomes from clinically relevant bacterial strains as input.

Identification and ranking: EDEN then identifies unique feature combinations and predicts previously untested proteins, scoring each of them from 0 to 1 for their probability of eliciting a protective immune response.

Antigen optimization: The output is a list of novel, protective candidate antigens that, based on antigen optimization strategies incorporated in EDEN, are optimized in terms of antigenic and structural properties as well as ease of production.

Verification: Top-ranking antigens are verified in pre-clinical models and assays and further optimized to derive an optimal, potent vaccine combination.

• EDEN’s ability to predict protective vaccine antigens has been shown in pre-clinical models. Once clinically validated, we believe our approach may have the ability to improve the attrition rates for new vaccine product candidates.
• While traditional reverse vaccinology (RV) relies heavily on previously tested antigens, EDEN is trained to identify underlying feature patterns that enable the discovery of novel and unbiased targets.
• With carefully curated data, EDEN has learned to filter out irrelevant proteins, narrowing the field of candidates from thousands to a few dozen proteins and reducing the burden on pre-clinical development.
• The rapid “evolution” of the genome that can occur in some bacterial pathogens makes it difficult to capture all pathogen strains with a single vaccine. EDEN is capable of leveraging genomic sequencing data to find important targets or domains that are present in the majority of clinical strains. By combining the correct antigens, EDEN can provide broad protection and cover most, if not all, relevant strains.
• Developing and verifying the safety and efficacy of a novel vaccine usually takes between 10 and 15 years, often resulting in a new vaccine arriving too late on the market to influence the spread of infections in the general population. In contrast, EDEN is capable of identifying vaccine antigens in a matter of weeks instead of years, potentially lowering the overall development time significantly.
• EDEN is highly scalable because it can rapidly identify a broad range of vaccine antigens against almost any bacteria, including drug-resistant bacteria, such as MRSA.