The models are retrained on an ad hoc basis. The models we provide are intended to cover as much of the chemical space as possible given the available data and as such the addition of extra compounds is less likely to improve the model’s applicability but may improve its accuracy. As we are dependent on public data from consistent assay protocols, the amount of new data each year is small.

We are constantly evaluating the available data and exploring new models. Please check StarDrop release notes regularly for model updates.

More StarDrop resources