RAVEN™ The ability to target any viral pathogen in any population group

Our RAVEN™ (Rapidly Adaptive Viral Response) model builds on the key features of PIONEER™ and EDEN™, together with novel, in-house developed evolutionary and structural capabilities. The edge that RAVEN™ has over other state-of-the-art methods is that in addition to identifying viral B-cell antigens, RAVEN™ also identifies multiple T-cell epitopes. This means RAVEN-designed vaccines are able to elicit a robust and durable antibody and cellular response – implying a broader, more reliable and more future variant-agnostic vaccine.

​​RAVEN™ can also be used to create a tailored response against selected viral variants or human populations, either for specific or broad coverage. Additionally, the rapid prediction of the machine learning components also allows for an unprecedented turn-around speed for novel design for a given viral strain.

Prediction: Using genomic and proteomic data from any virus target, together with MCH data, RAVEN™ predicts cross-protective B-cell antigens and protective T-cell epitopes.

Protein ranking: B-cell antigens are identified using a proprietary machine-learning ensemble adapted from our EDEN™ model, which interprets immunologically-relevant information about bacterial antigens that induce humoral protection in a vaccine setting. Through protein ranking, RAVEN™ selects verified antigens that would target the majority of pathogen strains with a single vaccine. 

Broad yet targeted: With a proprietary model composed of several computational features adapted from our PIONEER™ model that identifies peptides on MHC class I and II molecules, RAVEN™ selects protective T-cell epitopes. Working with 7 million parameters and a dataset of 1.4 million combinations, RAVEN™ downselects the T-cell epitopes to a set of peptides containing epitopes that would cover the desired virus population, whether that means selected viral strains in outbreaks, specific human populations, or single individuals.

Robust protection: The identified B- and T-cell antigens can be administered as individual components, but RAVEN™ is also able to further analyze the B-cell antigen sequence to identify sites where T-cell epitopes can be grafted into the protein. This produces a T-cell response while preserving B-cell antibody generation, which our data shows results in increased breadth, avidity, magnitude, and duration of the immune response. These identified antigens can be administered by any established vaccine delivery technology.

• RAVEN’s™ AI components enable the identification and combination of T-cell epitopes that cover the entire immunological diversity of the human populations (HLA type).
• By combining multiple potent epitopes in one vaccine, different T-cells are able to target the same infected cell, curtailing the spread of the infection more effectively.
• Selecting epitopes from all proteins in the viral genome means that vaccine-generated T-cells are able to kill infected cells at any stage of the viral replication cycle.
• Combining multiple epitopes ensures that any given variant of a strain is covered by conserved T-cell epitopes, meaning new mutations are likely to have little negative impact on the vaccine efficacy.
• Minimal constructs are designed from viral fusion proteins for the generation of neutralizing antibodies.
• RAVEN’s™ viral fusion protein antigen is designed using information from all available variants of the target strain to ensure that the generated antibodies offer cross-reactive neutralization.
• The RAVEN™ model can be applied to any known virus and is delivery model agnostic.