"Working with the supercomputer is complex because you work with different programmes, you are given a limited 'time budget' and have to specify exactly what the supercomputer should work on and how much memory is needed to do so."
Case study
Treating industrial wastewater with bacteria
How do you process an endless amount of data on DNA material from micro-organisms? During her research on purifying industrial wastewater, Marjet Oosterkamp (TU Delft) turned to the national supercomputer, which takes over the thinking when the capacity of the human brain and an ordinary computer are reached.
In short
Who: Marjet Oosterkamp
Position: Researcher
Organisation: Delft University of Technology (TU Delft)
Service: National Supercomputer
Challenge: Analysis of huge amounts of DNA data from micro-organisms for research into purification of industrial wastewater.
Solution: By using the national supercomputer, the data could be processed in a short time, providing insight into which micro-organisms best purify wastewater under which conditions.
You put in a soupçon of data, the supercomputer goes to work on it and a few days later delivers bite-sized chunks that you can puzzle out further. It sounds simple, but Oosterkamp says it took quite some time and effort to get to grips with the supercomputer.
"It's complex because you work with different programmes, you are given a limited 'time budget' and you have to specify exactly what the supercomputer has to work with and how much memory it needs. Fortunately, we were helped by our faculty's ict manager, a bioinformatician from Mexico and, of course, the experts at SURF, the organisation that manages the supercomputer."
Optimal conditions for water purification
Oosterkamp is conducting research with a team of scientists under the project name BioXtreme to find a method to purify industrial wastewater with micro-organisms such as bacteria and archaea. In about four membrane bioreactors in the Delft lab, they use model wastewater to create conditions in which the micro-organisms purify difficult-to-degrade substances such as phenols from the water.
"We use model wastewater with a recipe of components commonly found in industry," says Oosterkamp. "That way we know exactly what is in the water and can analyse everything properly. Eventually, we do want to work with real wastewater from industry, but that is a next step."
The aim of the study is to find out at which conditions (temperature, salinity and lactic acid content) the micro-organisms can do their job optimally. Indeed, according to Oosterkamp, much is still unknown about the role of high temperatures in wastewater treatment by microorganisms. What remains after treatment are water that can be reused in industry and biogases that can serve as a source of energy. A circular system, in other words.
From samples to calculated data
Samples are taken from the model wastewater from which DNA material from micro-organisms is isolated, Oosterkamp explains. "We send that to a company in Leiden, BaseClear, which breaks up all the material into very small pieces in their laboratory. We get that huge mess of data back."
From her computer in Delft, she sends all that data to SURF's data centre, located in a 72-metre-high building on the Science Park in Amsterdam. There, spread over two floors, the data centre includes 100 kilometres of network cable, 25 petabytes (25,000 terabytes) of online storage and 50 petabytes of tape storage. On SURF's server system, several processors all perform a particular task. They communicate the results to the central server, which reduces it back to a set of calculated data.
Impressive to see, Oosterkamp found when she visited the data centre earlier this year. "There are huge computer cabinets where the servers run and ingenious systems for air regulation. It's the latest of the latest, you're not allowed to just go in there either."
"We use model wastewater with a recipe of components commonly found in industry. That way, we know exactly what is in the water and can analyse everything properly."
Further puzzling
For the project, the researchers needed 1 terabyte of working memory from SURF. The supercomputer then took 400 hours (almost 17 days) to perform the computational work. Now that the work at SURF is over, the researchers can continue their work using 'normal' computers. "We are in the final phase of the project. It is now a matter of puzzling and figuring out which DNA belongs to which micro-organisms, so that we know exactly what properties they possess."
"A regular computer simply cannot handle that much data or it will take months before you get some results"
According to Oosterkamp, the help of the supercomputer was essential to do the research at all. "An ordinary computer simply cannot handle that much data or it takes months before you get some results. That's far too long, especially since you don't know beforehand whether it produces the desired results and you might have to start all over again."
The results of this project can in turn form a basis for the next project, where the same test will be carried out, for example, on real industrial wastewater. "For that, we will definitely need SURF's supercomputer again. It was very challenging and interesting to work with SURF. I am already looking forward to the next 'collaboration'."
This article was previously published on the TU Delft website.
Team Data & Computing Services
E-mail