The talk, called “Design for Inference in Medicines,” focused on how researchers can integrate computational modelling and experiments to generate the data needed to develop new medicines.
“To really leverage technology to make real medicines it’s not enough to have great techniques; you have to combine it with great biological insights,” she said. “While all of these medicines have brought great benefits, there’s still a lot of unmet need in everything from infectious disease to cancer to neurodegeneration. [Genentech’s] ambition is to try to double the amount of medicines we make and try to do it at a lesser cost to society.”
Regev says achieving that goal can start with human genetics. The field has identified 100,000 regions in the human genome where variants are associated with diseases. The problem, Regev says, is predicting what all these variants do and how they do it – a difficult and laborious process.
This is where advanced technological techniques such as computational biology, machine learning and single cell RNA sequencing can come in, she said, “The question is can we design experiments and algorithms in new ways that would allow us to make better inferences of mechanisms and medicines in patients.”
Much of Regev’s talk focused on the “lab-in-a-loop,” framework that Genentech uses to create a feedback loop, with the data generated in experiments being fed into a computational model, which is constantly refining the experiments.