Martin-Immanuel Bittner

Arctoris, UK

Title: Implementing fully automated kinase inhibitor characterization for AI-driven drug discovery using a robotic system

---

Biography

I am the CEO of Arctoris, the world's first fully automated drug discovery platform that I co-founded in Oxford in 2016. My background is in medicine and cancer research, having completed my MD at the University of Freiburg, and my DPhil as a Rhodes Scholar at the University of Oxford.

Abstract

About 50 kinase inhibitors have been approved by the FDA for different indications so far. On the way from target validation to approval, the biochemical characterisation of a novel kinase inhibitor is a tedious, yet absolutely critical task. High resolution, kinetic molecular profiling can enable better data-driven decision making early on in the drug discovery process, not only saving time and resources, but also leading to superior molecular design – especially when combining human with machine intelligence.

Arctoris developed a robotics-enabled process for fully automated kinase inhibitor characterisation on its Ulysses technology platform, providing an unparalleled depth of data capture, going beyond the current state-of-the-art of biochemical assay setup. We validated our technology platform establishing assays against four members of the Janus Kinase family (JAK1, JAK2, JAK3, TYK2), profiling a set of JAK inhibitors. Of note, several JAK inhibitors with prior FDA approval for other indications entered clinical trials for COVID-19 treatment, making this target class particularly relevant for an in-depth study.

Reagent validation, assay development, calibration, and optimisation were expedited through systematic multifactorial experimental design, high density assay plate formats and versatile automated liquid handling. Fully automated protocols were optimised, validated, versioned, and explicitly encoded.

Robust potency measurements of all inhibitors were established against each of the JAK targets, enabling the identification of molecules within the JAK inhibitor set that exhibit a range of kinetic properties.

Our in-depth biochemical characterisation and profiling process generates more than 100x more data per assay compared to conventional manual laboratory operations, leading to significant improvements in data generation and data capture. These improvements are of particular importance for AI-driven drug discovery programmes, where access to structured, reproducible, and well annotated, data is of critical importance for model training and validation.