Cortex Virtual HTS

Drastically reduce the size, cost and time of your screening campaigns!

Our AI learns from HTS data and can be used to virtually screen any libraries of your choice or all publicly available libraries, up to billions of compounds. You get a qualified hit list annotated with ADMET and other selected relevant properties, such as fluorescence, specific toxicity, etc.

Since we just need your activity scores and the molecules’ SMILES, we offer the options to train from your datasets without knowing your target or property of interest. Alternatively, if you have no or little data, or if you cannot disclose it, we can train from our data only, providing that the target is disclosed and available in our database.

Hit selection workflow

Here is a concrete example of how our service can be integrated in a hit selection workflow, leading to more experimentally qualified hits at a fraction of the total cost. These numbers are based on external in vitro validations of our predictions. Our service is similarly useful for small screens as well, down to ~1000 compounds.

Accuracy is everything

Cortex specializes in making accurate predictions, quantified here using ROC-AUC: a metric which is robust to differences in scoring system, choice of decision threshold, or test set composition.

Our AI's prediction accuracies are on a par with experiments, with AUCs typically between 90% and 99% (measured on held-out data or external dataset). Cortex's predicted hits are truly novel and not simply existing hit analogs.

Each increment in accuracy translates to manifold more real hits


Our pricing model is attractive for both small and large projects. We will provide a quote tailored to your needs, based on the number of compounds to be screened and the targets or properties to be tested.

In every case, the aim of our service is to help you obtain an increased number and quality of hits at a dramatically reduced cost compared to relying purely on in vitro experiments.

Contact Us