ADONIS - Autonomous Drug Combination Screening using Chemical and Genetic Perturbations

Drug combinations constitute an important therapeutic strategy, but the huge combinatorial space makes screening infeasible using conventional methods.

The purpose of this project is to develop a novel phenotypic screening platform (ADONIS) that can autonomously explore and identify efficient drug combinations.We will develop iterative screening for drug combinations that shift cells towards a desired state.

We use multiplexed microscopy imaging (Cell Painting) as profiling technology, which constitutes an excellent tradeoff between cost, speed, and information content.

In our previous VR-funded project we developed a robotized lab and implemented fully automated Cell Painting, and in this project we will expand it to perform: i) autonomous screening for drug combinations towards soft-tissue sarcoma and colorectal cancer; ii) genetic knock-out experiments to illuminate drug resistance; and iii) generative AI to design novel drug leads.Our approach stands out by using image-based cell profiling data that allows for using the latest deep learning methods, and an AI model that selects the next batch of experiments to be carried out in an automated lab.

This makes it possible to convert brute-force approaches, such as drug screening, into search problems that can be approached systematically with machine intelligence.

With our project we will contribute to realizing this transition to more efficient scientific discoveries where AI iteratively interacts with the physical world.