A fast and energy-efficient way to compute the most likely mutations in a cancer tumor

What is it about?

Modern DNA analysis methods make errors. There is an algorithm called SCITE that rectifies some of these errors and finds out which mutations are most likely to be present in a cancer tumor. We have re-implemented the SCITE algorithm with special computation hardware, so-called Field-Programmable Gate Arrays (FPGAs), to make this evaluation faster and more energy-efficient. In our work, we describe how we designed and optimized our implementation and how its speed and energy efficiency are comparable and even higher than what traditional computers can achieve.

Featured Image

Photo by Patrik Kernstock on Unsplash

Photo by Patrik Kernstock on Unsplash

Why is it important?

This work is another example of how FPGAs can be used to accelerate scientific computations at much higher performance and energy efficiency than what traditional computers can achieve. This is especially true since traditional computers are optimized for a certain set of operations, e.g. floating-point arithmetics. This means that their performance quickly falls when an algorithm like SCITE uses operations not within this set. FPGAs can execute almost any operation equally well. Thus, FPGAs can shine in situations like this where traditional computers fail.