Not all scientists wear lab coats: Inside the digital engine room of HYDROCOW
When people think of biotech research, they often picture pipettes, petri dishes, and lab coats. But in the Behind the Hydrocow series, we’re here to show you that not all breakthroughs happen at the lab bench. In this post, Samira van den Bogaard takes us into the world of computational biology—where code meets cells, and where a powerful model can save months of lab work. Welcome to the other side of HYDROCOW, where microbes are engineered not in test tubes, but in lines of code.
A cup of tea or coffee, a comfy desk, and a powerful (enough) computer, that is all we, at the Industrial Systems Biotechnology group in Aachen, need for our work on the HYDROCOW project. Most of the time, when I tell people about the project, they expect that we run around in lab coats doing cool sciency stuff with SoF1. But when I tell them we only do computational work in Aachen, they cannot paint a picture in their mind of what we are actually doing. Many people understand that you need many experiments to build a milk protein cell factory, but what they are commonly not aware of is that computational methods can decrease this number of experiments. But how?
“Many people understand that you need many experiments to build a milk protein cell factory, but what they are commonly not aware of is that computational methods can decrease this number of experiments. But how?”
Imagine: SoF1 has 4520 genes. Most genes can be translated into enzymes, each having a different function in the cell. The cell is a like chemical factory: to function and produce more biomass, it converts nutrients into building blocks using branched chains of chemical reactions. Like a factory, these chemical reactions are sped up by ‘machines’, which in biological systems are the enzymes. However, while most factories focus on max 10 chemical conversions over a timespan of seconds to days, cells perform thousands in microseconds! You can imagine that with all the ‘machines’ to make this happen, the cell is a crowded place. And then the investment in making the ‘machines’: roughly 50% of the cellular energy goes to the production of enzymes. It is therefore important for a cell to use its resources wisely and optimally, and most microbes have evolved over millions of years to achieve such a delicate balance. And the cool thing is, if we know the genome and on which food a microbe can grow, we can express all this information in mathematical models, which is what Aziz is doing.
Okay, you’ve learned about how complex cells are and that we can capture this in models, what does this have to do with the HYDROCOW project? First, within this complex network of chemical reactions, genes, and enzymes, we need to determine which targets to test experimentally. We can use information from databases to compare SoF1 to other, similar, microbes with some desired behavior. For example, we can see which of these relatives to SoF1 grows on a specific nutrient and use this information to adapt SoF1 to grow on this nutrient as well. Karan from our team, works on this ‘pangenome’ analysis for different targets.
Secondly, we can use clever optimization techniques to entangle the complex network of chemical reactions and find those places where we can incorporate the milk production pathway, and which genes are associated with it. Aziz will use his model to design these cell factories. Thirdly, we can leverage more complex models to determine how the allocation of resources changes when we do these modifications. As mentioned before, cells are optimized to do what they do for millions of years, and when we make them use some valuable resources to produce some molecule which is useless to them, they are not happy. You could imagine, if I asked you to spend 50% of all your money on a useless toy for someone else, you would also want to find a way out of the deal. This also happens to microbes: they can kick out their production pathway, stop growing, or even die. By using information about the efficiency with which enzymes catalyze reactions, we can extend the purely biochemical model into a more complex model, yielding information about the changes if we do these modifications, and predictions on how we can tune our production pathway such that it doesn’t bother SoF1 that much. Establishing these methods and building the software to support them is my daily business.
So, now you know what we computational people do all day at our desk. We study microbes, we engineer, we design, and of course we write code. What we do is a lot of fun, so if you’re interested, join the dark side!
Happy coding!
Samira van den Bogaard
PhD Student in metabolic modeling, RWTH Aachen University