Products
Off-target toxicity prediction
Off-target toxicity prediction is a crucial aspect of preclinical evaluation for T cell and antibody therapies, as unintended interactions with epitopes present in normal tissues can lead to severe adverse effects. We have developed a sophisticated model to score binding similarities between therapeutic targets and other peptides based on provided X-scan (or Ala-scan) data.
Using this binder specific model, we score binding similarity for all the peptides within our extensive database, which includes an impressive collection 1.4 million unique peptides from of 4,200 different samples. Beyond evaluating binding similarity, our model integrates the HLA binding profiles of the peptides, adding a crucial layer of information that enhances the accuracy of off-target predictions. Additionally, our database provides robust evidence for the presence of these epitopes in various healthy tissues and organs, facilitating a more thorough assessment of potential off-target effects. By leveraging this extensive dataset, our model significantly improves the reliability of preclinical evaluations, ensuring safer and more effective therapeutic development.
If you are interested in off-target toxicity prediction, please contact us below
Contact us
Figure 1: Overview on the workflow to identify potential off-targets from TCR affinity data. A) TCR affinity data (e.g. x-scan) is transformed into a ML model predicting TCR-HLA peptide binding. B) After submitting all HLA peptides present in healthy organs to the ML model, a score-cutoff for potential binding is calculated. C) Sequence motif from peptides that are predicted to bind to the TCR. D) List of peptides and safety relevant data such as expression in healthy organs and healthy primary cells, TCR binding score and number of healthy samples that are positive for the given peptide.