Accessible, low-cost super computing power

As a specialist in light and electron microscopy, cellular biologist Jack Fransen (Radboudumc), uses the computing power of Azure DRE to analyze the images he produces with his equipment. Due to the large format of the microscopic images, this is, in and of itself, a huge stride forward: making it simple to analyze data while cutting costs.

Celluar biologist Jack Fransen is a manager at Radboudumc’s Technology Center Microscopy. He started working with the Digital Research Environment (Azure DRE) when it was first launched in 2017, but at the start, it could not serve his needs. “The images are huge files. I had to physically send data on a hard drive to collaborating researchers. To give you an idea of how large: I remember traveling with a hard drive that contained three images – a terabyte each – from the United States.”

“You’re talking about computing power to be able to process some 100,000 images. The data in these images need to be averaged and fitted using special algorithms on a pixel-by-pixel basis to calculate one final image that we can work with. On a high-end PC, this can take up hours to days.”

New possibilities

"The first version of Azure DRE did not have the computing power to process these microscopic images. I'm enthusiastic about the new possibilities of Azure DRE 2.0. For the past five months or so, we have been using DRE to analyze data on different virtual machines in parallel, and the results are very promising."

Jack Fransen explains that everyone in his field is looking for faster and cheaper ways to process the huge amount of data they collect. “Some universities are looking for answers in creating a large cluster of local computers that could deliver the necessary computing power. This has a few disadvantages. Firstly, the data is only accessible in one physical place. Secondly, when researchers use the machines for analysis, it cannot be used for research. Thirdly, these machines are very costly in both purchase and maintenance.”

“The virtual machine, on the other hand, has none of these disadvantages. The data can be accessed at any time and in any place. Researchers needing to analyze their data are not holding up our equipment and thus hindering new research. And the difference in costs is substantial. Within DRE, you only pay for computing power when you're using it. In other words, you can temporarily up the computing capacity. At € 1,50 an hour, it's easy to calculate that you are better off using a virtual platform than expensive machines that could cost anything up to € 11,000 a piece.”

Storing and sharing

“Of course, the Azure DRE environment is not yet perfect,” says Jack. “At the moment, it does take some time to upload the data to the cloud. The speed will need to improve, but I’m confident that this is where the future is, rather than physical computers.”

It is also good to note that, unlike the data derived in other fields of research, Jack Fransen and his colleagues don’t use DRE as a permanent storage database. The cost of keeping the files of a microscopic image in the cloud is too much. For now, images will still need to be stored on a physical computer.

“There are national projects in the field that are trying to find cheap alternatives to store and share data in a cloud-based environment,” says Jack. That’s not to say that Azure DRE will be obsolete when a possible other sharing environment is created. According to Jack, Azure DRE offers a great workspace to help researchers with their analysis: “Users can customize the Azure DRE workspace with any available open-source software that suits our needs. We can also add analysis protocols to ensure that researchers are working the same way. It’s also a user-friendly environment, meaning that even researchers that are less computer savvy can use it instinctively. A real plus for researchers who want to focus on research rather than digital tools.”

Want more information about Azure DRE and the possibilities for your organization? Read on