Where do systems biology and bioinformatics come together?
Bas Teusink: All biology is a product of evolution – and that leads to certain observable designs. As systems biologists, we essentially reverse engineer these designs to understand how their different components work, and how systems behave based on how molecules interact.
Jaap Heringa: In bioinformatics, on the other hand, we view organisms as information-processing systems. Molecular systems carry information and tell us stories – we try to understand this informational system. We label all the data, dissecting DNA into its constituent parts for example, so that we can start to combine data sets and create models using computational methods.
BT: You can think of it like a game of Lego. The bioinformatics experts categorise all the Lego blocks into different piles based on comparison with other games, and label them. The systems biologists then build a model using those blocks.
How does all this apply in the real world?
BT: We help biologists to deal with high throughput data – so we work in areas like metabolomics (how cell metabolism works) and proteomics (how protein networks work). When it comes to disease in patients, you take a blood sample and measure the data you get from the sample to try to understand what’s different in the patient. The type of modelling that we do can help understand disease mechanisms, or can help with drug targeting or personalised medicine.
I’m currently more active in applying metabolic models to microbiology and biotechnology: using data to improve fermented foods or understand how metabolic networks can be rewired to replace fossil fuels with energy made from plant-based waste. But I’ve also used these models to better understand the metabolism of cell cultures.
JH: Another example from the health domain is Covid-19. We can look at the data sequence of the virus and compare it with other viruses. Which family is it a part of? How could it relate to the structure of the proteins encoded by the virus? Alongside a toolbox of bioinformatics analysis methods, this data help us understand the virus.
What’s in the toolbox?
JH: We create different methods in order to make better comparisons. So we primarily develop and use computational methods to learn from comparing DNA, RNA and protein data, based on evolutionary and biological principles. But we’re also working on other modelling techniques to reach the right level of detail for the research in question.
How does this computational modelling contribute to animal-free research?
JH: There’s a huge amount of data out there, but the challenge is to make it accessible and comparable for different purposes. For example, if a scientist in the Netherlands is thinking about testing a particular substance on a mouse, but a scientist in Japan has already done that same test, then in theory that data is already out there. There’s no need to repeat the test so long as the scientist in the Netherlands can access the data – which therefore reduces the amount of animal testing needed. We can even go one step further: there’s no need to use a mouse model at all if the data model is better.
BT: There’s a general consensus that animals are often not particularly good models for humans. Whereas by combining data, you might apply modelling to see how well a mouse model translates to a patient, or whether there are crucial differences between the two. There is huge potential here: for example, I know of a study in which a mechanistic model of hormone signalling –based on cell lines – was used to identify the critical biomarkers that predict the responses of patients to treatment. In this way, cellular models can be much more useful than animal models in predicting responses to a drug in a patient. This is where the concept of a “digital twin” comes in.
What’s a digital twin?
JH: Essentially, if you have enough information about a real-life organism, you can create a digital model that is the twin of the original. A digital twin is a nice buzzword for a mechanistic replica. It’s mainstream in engineering, but it’s certainly not mainstream in biology yet. Data science and modelling is a huge topic in other areas of life, but it’s not yet being sufficiently applied in biomedical research. There are a few good proofs of concept already, but I think digital modelling in our area will take another five to ten years to become mainstream.
And how about the potential of data and models to replace animal research?
BT: We’re not there yet. For more complex behaviour in particular, such as that produced by the brain, it’s difficult to test responses without animal models. The key questions in these cases are: How can we translate the data from animal models to the human situation? How can we integrate data from animal models, cellular models and patient models? At this point, it’s more about enhancing our understanding and interpretation of data from animal models than replacing them altogether. But in many other cases, there is clearly the potential to leave animal models out of the equation.
What are your priorities in this field over the next five years?
JH: One key problem that we’re working on is ensuring FAIR data – by which we mean data that is findable, accessible, interoperable and reusable. The data infrastructure needs to be developed in such a way that data is available and understandable for people and computers to be able to use it for new purposes. There’s also the translational aspect: we need to encourage scientists who are doing animal testing to link up with systems biologists so that they’re thinking at this holistic level. It’s about building the entire system.
BT: Education and awareness are also key. We need to make people aware of what’s out there and how they can access data in the FAIR way. And we need to educate the next generation of biologists so that they’re more focused on data and computational modelling, and better able to use them as tools. We collaborate very intensively on teaching already – it’s important to reach students early on.
Speaking of collaboration, that’s also important. I work with cancer biologists on the metabolism of cancer cells, for example. And Jaap works with the Dutch Cancer Institute (NKI) on cancer research that relates to genes or DNA. Another example is Covid-19: we think there may be a metabolic component to long Covid, so we’re trying to get funding to collaborate here as well.
Where can people go to find more information?
- ELIXIR unites Europe’s leading life science organisations in managing and safeguarding the increasing volume of data being generated by publicly funded research. It coordinates, integrates and sustains bioinformatics resources across its member states and enables users in academia and industry to access services that are vital for their research.
- The Dutch Techcentre for Life Sciences is a national platform that connects experts in digital life sciences and enables easy access to relevant data and tools. It supports advanced computational analysis and modelling via bioinformatics and computational modelling techniques, including machine learning and artificial intelligence.
- The Netherlands Bioinformatics and Systems Biology (BioSB) research school is a national platform for the Dutch community of researchers in bioinformatics and systems biology.
- The Amsterdam Institute of Molecular and Life Sciences (AIMMS) focuses on achieving breakthroughs in molecular, pharmaceutical and life sciences, through fundamental understanding of molecules and their interaction with biological systems, humans and the environment.
Interview by Vicky Hampton, December 2021