The Cray XC30 super computer helps build the human genome in 9 minutes

With the help of supercomputers, scientists were able to shorten the time to build a biological genome very much.

Super computer builds human genome in 9 minutes

A group of members at the Joint Genome Institute - affiliated with Lawrence Barkeley National Laboratory , and researchers at the University of California Barkeley recently used 15,000 employees of the Cray XC30 ' Edison ' supercomputer to shorten work. Build the entire human genome. The total time for Edison supercomputers to do this was 8 minutes 24 seconds.

Current supercomputers play a rather important role in shortening the time to conduct large-scale scientific research. For genetic research , this achievement opens up more possibilities than genome building with ' de novo ' assembly tools like before. For example, with the 'de novo ' tool, the Meraculous code break takes nearly 24 hours. The complete genome building of wheat plants - which is extremely complex - is almost beyond the capabilities of this tool. But with the HipMer method , under the help of a 15,000-core supercomputer, the wheat genome was built in less than 40 minutes.

Supercomputers increasingly play an important role in scientific research

The team said, although the 'de novo' method has the advantage of ' finding new gene combinations that may have been missed', the time it takes to do the job is big again. far more than modern assembly tools.

Newly applied method called HipMer (short for High performance Meraculous) can help scientists solve the previous problem - Handling a large amount of data and solving extremely complex algorithms trash.

'For the first time in history, we have been able to overcome the capabilities of the entire set of genetic engineering devices in the world, and open a new era for genome research. The new method combines the determination of high efficiency sequences, while also obtaining the details and completeness of the de novo method. In this way, we believe it will make money for future developments in biomedical technology, including food security and the preparation of drugs for individuals. '

Edison supercomputers were designed a year ago to help scientists perform genome studies. Therefore, ' Edison ' is capable of balancing the processing of a large amount of data, as well as meeting the tasks that require multiple processor cores. When NERSC members worked with Cray to develop the XC30 used for research, in addition to the ability to link a large number of processors, they also wanted a memory with high bandwidth traffic. As a result, a supercomputer with the speed of accessing data is about 163 gigabytes per second and falls in about 2.4 petaflops.

Time to build a genome using HipMer method

For supercomputers, processors and memory bandwidth are two important points. However, the construction of the code set to be able to distribute the work efficiency between 15,000 Ivy Bridge processors 12 individuals is equally important. The Edison supercomputer was built on the basis of the Meraculous genomic tool - now considered the best tool using the " de novo" method.

The HipMer method is as advanced as using a variety of algorithms to optimize UPC performance, improve access speed, as well as perform complex operations simultaneously. At the same time, reading and writing data from a large number simultaneously from microprocessors is also done better.

The team is also developing another method of memory allocation for HipMer, which when successful will shorten the full genome of wheat in less than 11 minutes. The next phase of the research team is to apply this method to genome research on a larger scale.