Alumnus Robert Ietswaart: ‘Machine learning is revolutionising drug discovery’
Robert Ietswaart does research into gene regulation at the famous Harvard Medical School in Boston. He developed an algorithm to better predict whether a candidate medicine is going to produce side effects. He studied mathematics and physics in Leiden, and gained his PhD in computational biology in the United Kingdom.
One thing Ietswaart still appreciates from his time as a student at Leiden is that he was able to develop as he wanted to. He liked modelling molecules in a cell, and was able to select subjects to go with that. ‘I did my bachelor’s research in the field of biophysics’, he explains, by way of a Teams connection with Boston. ‘In my master’s, I mainly studied theoretical physics in depth, from biophysics to string theory. And I’m still grateful that all that was made possible. I really enjoyed learning a lot but I was also a member of Catena, where I partied hard too.’
To Boston
In his master’s research, Ietswaart leaned more towards biophysics. His supervisors were professors Helmut Schiessel of the Lorentz Instituut and Martin Howard of the British John Innes Centre in Norwich. Supported by grants from the LUF, the European Erasmus Programme and the university’s ‘Outbound’ programme, he was able to do part of his research in the UK. And once more supported by grants (the Prins Bernhard Cultuurfonds and the VSBfonds), he was also able to start his PhD research at the John Innes Centre: computational biology, in particular in the field of gene regulation in plants. Ietswaart came in contact with Professor Stirling Chapman of Harvard Medical School in Boston and she wanted him in her team. So off it was to Boston. Ietswaart: ‘As a postdoctoral student at Harvard, I’m free to research what I think it important. That comes with a hefty responsibility to get good results.’
Who: Robert Ietswaart (33)
What: double bachelor’s in Mathematics and Physics (2009 - cum laude) and Master’s in Theoretical Physics (2011 – cum laude). Was awarded the prize for physics talent by Casimir Onderzoeksschool for Natuurkunde (research school of physics) in 2010. Obtained his PhD at the John Innes Centre in Norwich, UK. Has worked at Harvard Medical School in Boston, US, since 2016.
Association: Catena
Best spot in Leiden: ‘I always thought De Burcht was fantastic and I also really liked Marekerk.’
People and plants are not that different at a basic level
The move to human gene regulation in Boston that Ietswaart is now working on was not such a big one, he says. ‘You’re doing fundamental research at molecular level and the processes between plants and people don’t differ so much at that level. With the help of machine learning (ML), we’re trying to get a better understanding of gene regulation on the scale of the entire human genome. The hope is that we will be able to select proteins that can serve as molecular goals for new medicines for diseases such as lung cancer.’ Ietswaart has now developed two ML techniques. Working with researchers from Novartis pharmaceuticals, he has made an algorithm to better predict whether a candidate medicine will produce side effects. This means that the effectiveness of a medicine could be established in a much earlier phase. Ietswaart: ‘I developed the other technique to achieve a better understanding of what proteins do in a cell, for example if a medicine is administered. In the future, I want to further expand my research into ML in gene regulation for the discovery of medicines.’
Robert Ietswaart on the current state of affairs in medicine research: ‘There’s a real revolution in the discovery of medicines at present: ML makes it possible to systematically use large biometric datasets to find patterns and consequently discover candidate medicines faster and less expensively. We have roughly 20,000 genes in our genome that code for a protein (determine which process that protein initiates and which it doesn't). Those proteins can interact with each other and when they do, a large network of molecules is created that joins to regulate those cellular processes. For example, learning which specific proteins and interactions are important for the fast cell division of a cancer cell is a difficult task. ML is an important new technique with which to tackle this problem. In my work, I make grateful use of the quantitative techniques I learned at Leiden during my studies.’
Photo: De Harvard Medical School
To Europe?
Working on his PhD in the UK got Ietswaart more than just a degree: he met his wife Ly there, a Vietnamese master’s student. She was studying Brand Leadership at Norwich Business School. Ietswaart travelled to Vietnam to ask her parents for her hand in marriage – and was given their blessing. In the little spare time left outside research work, the couple like to take walks on the beach or outdoors in the Boston area. Despite the fact that Boston has a wonderful ecosystem of scientific institutes and businesses that are willing to collaborate with each other, Ietswaart says that he and his wife are seriously considering returning to Europe. The postdoctoral period is coming to an end and Ietswaart would also like to move closer to his family and the lifelong friends he made as a student. In addition, housing costs have risen rapidly in Boston; affordable accommodation is hard to find. Unfortunately, we can’t reassure him on that point as far as Europe, and in particular the Netherlands, is concerned.
Why new medicines are expensive
There’s another subject Ietswaart wants to discuss and that’s the Dutch/European attitude towards medicine prices: ‘It currently costs a pharmaceutical company an average of two billion euros to develop a new medicine and bring it to the market. In Europe and the US, a company gets a patent of 20 years on a new medicine. In that period, the company has to both recoup the costs and invest money in more new medicine development.’ What Ietswaart wants to say is that without the initial high-price phase, pharmaceutical companies would no longer be viable, the consequence of that being no new medicines and continued suffering for patients.
‘I would love it if new public-private initiatives were to come into being, so that we could cover the costs of medicine development without being so dependent on the pharmaceuticals industry, often American. Then new medicines would enter the European and Dutch markets faster and be more affordable.’ But according to Ietswaart, the bottom line is: without investment in research and development, there will be no new medicines on the market. And if you want to pay less for more, you'll end up at the end of the queue. ‘We saw that with the COVID-19 vaccines’, says Ietswaart. ‘And that means it’s later before you get out of the crisis. We need to be better prepared for the future.’
Recent publications by Robert Ietswaart:
- Machine learning guided association of adverse drug reactions with in vitro target-based pharmacology
- GeneWalk identifies relevant gene functions for a biological context using network representation learning
Text: Corine Hendriks
Mail the editor