It provides a set of compounds by analyzing and selecting potential hit molecules from large databases or even scratch.
It optimizes molecules from the first hits with/without structural similarity by quantitative structure-activity relationship and reinforcement learning.
It predicts ADME-T characteristics of small molecules and suggests new derivatives with better ADME-T profiles than original.
It provides a set of variants for optimizing enzymatic activity, binding affinity, selectivity, protein expression, thermal stability, proteolytic stability, aggregation, and so on.
It optimizes proteins for better activity and thermal stability by quantitative structure-activity relationship and machine learning.
It predicts many properties of the target proteins such as proteolytic cleavage site, immunogenic site, aggregation site, nonspecficity of single-chain fragment variables, isoelectric points, and so on.