OncoNeo400 AI Engine – bioaiworks.com

OncoNeo400 combines AI mass predictions with Advanced Statistics and Biological reasoning to make more reliable Neoantigen Immunogenicity predictions

OncoNeo400 source of knowledge

The Onconeo400 was trained using various Real world experiments to be able to predict the binding affinity towards immune system cells

Most AI models rely on point estimates which are converted to classes using simple thresholds. However this approach doesn’t take into account the uncertainty of individual predictions. So when we say a model is 90% accurate we say this for the whole model, not for any individual prediction we made. Onco Neo400 can construct the confidence interval around every individual AI prediction using a swarm of 150 AI models. This means we can inject more certainty in predictions and make them more reliable and validated against experimental data.

The T cells not only have a role in recognition of Neoantigens on Cancer cell surface, but also actively engage with them, displaying cytotoxic effects and antigen presentation to other cells. T – Swarm is all about predicting these cellular effects using a swarm of AI models

The T swam part of the Engine is all about cellular reactions and i composed of 50 AI models able to predict if a T cell will recognize the Neoantigen on the surface of tumor cell as non-SELF but also which cellular reaction would it make.

The HLA swarm AI system can make predictions on 190+ HLA genotypes known to have a role in cancer immunity and carefully curated based on biological domain knowledge.

The B-SWARM. Another set of AI models with main characteristic or recognizing the BCR based receptor, but also the the reaction (antibody creation) that B cells would make in ex vivo presence of peptide. A good example of this would be serum with B cells in presence of the peptide.