Senolytic Drug Discovery as a Proving Ground for New Machine Learning Approaches
In principle, machine learning can be used to make small molecule drug discovery run more rapidly, more cost-effectively, and with a greater chance of success. The development of senolytic drugs to clear senescent cells is a good proving ground for this type of approach, and will likely accelerate investment into machine learning driven drug discovery platforms with broad application. Firstly, the state of the science shows that senescent cells are vulnerable to mechanisms that can be targeted effectively by small molecules. Secondly, it is also clear that far from all of these mechanisms are known and much remains to be discovered, as new approaches are emerging on a regular basis. Thirdly, the field is not well developed, yet potentially very large, with everyone much over the age of 50 as an intermittent patient. There is plenty of room to run a drug development program and achieve economic success, even given many other groups doing the same, and there will be many customers in the future for a marketplace of machine learning services. These incentives and predictions matter; they are necessary for a field to attract interest and grow.
Cellular senescence is a stress response involved in ageing and diverse disease processes including cancer, type-2 diabetes, osteoarthritis, and viral infection. Despite growing interest in targeted elimination of senescent cells, only few senolytics are known due to the lack of well-characterised molecular targets. Here, we report the discovery of three senolytics using cost-effective machine learning algorithms trained solely on published data. We computationally screened various chemical libraries and validated the senolytic action of ginkgetin, periplocin, and oleandrin in human cell lines under various modalities of senescence.
The compounds have potency comparable to known senolytics, and we show that oleandrin has improved potency over its target as compared to best-in-class alternatives. Our approach led to several hundred-fold reduction in drug screening costs and demonstrates that artificial intelligence can take maximum advantage of small and heterogeneous drug screening data, paving the way for new open science approaches to early-stage drug discovery.