DEARGEN -Model DearDTI -Drug Technology
DearDTI presents drug candidates with a good efficacy by predicting the binding affinity, based on information about the compound-protein interactions.
DearDTI predicts a binding affinity between compounds and proteins(Drug-Target Binding Affinity). It discovers small molecules with high potential as drug candidates by predicting that particular small molecule can be bound to which target or which small-molecule compounds will bind well to the desired disease target. DearDTI has made global context be used effectively by introducing Transformer technology based on the Self-Attention mechanism. Accordingly, DearDTI increases the precision of the model and suggests more accurate candidates by effectively reflecting the knowledge of compound learning patterns in the model.
The technology of predicting disease targets that bind to specific small molecule in the DearDTI enables drug development Pipeline for drug repurposing.DearDTI presents a drug development pipeline by predicting targets that candidates (fail to pass the clinical phase due to its low efficacy) can have a better effect or exploring new indications for existing approved pharmaceuticals.
The DearDTI model was published in the Journal of Machine Learning Research (JMLR) which is a journal with the highest level in artificial intelligence. In addition, it is presented at Machine Learning for Healthcare (MLHC) that is the largest conference on artificial intelligence technology using medical big data.