As humans, we each have trillions of cells. And every cell has a nucleus with individual genetic information –DNA – that may mutate to create an abnormality. If a human is born with an abundance of abnormalities inside cells, or if mutations develop over time, disease ensues. To make this much more complicated, cells are sometimes a mix of each abnormal and normal DNA – a mosaic, so to talk, and just like the art form, this complex montage is obscure. Nonetheless, a research team led by Joseph Gleeson, MD, Rady Professor of Neuroscience at UC San Diego School of Medicine and director of neuroscience research on the Rady Kid’s Institute for Genomic Medicine, has been using the Triton Shared Computing Cluster (TSCC) at San Diego Supercomputer Center (SDSC) at UC San Diego for data processing and model training to unveil recent methods for DNA mosaic recognition.

Gleeson and his team recently discovered recent genes and pathways within the malformation of cortical development, a spectrum of disorders that cause as much as 40 percent of drug-resistant focal epilepsy. Their research shows how computer-generated models can efficiently mimic human recognition work in a rather more efficient manner and was published this week in Nature Genetics. A related study was published earlier this month in Nature Biotechnology.

We began with a trial allocation on SDSC’s Comet supercomputer a few years ago and have been a part of the TSCC community for nearly a decade. TSCC allows us to plot models generated by a pc recognition program called DeepMosaic and these simulations allowed us to understand that when we trained the supercomputer program to discover abnormal areas of cells, we were capable of quickly examine 1000’s of mosaic variants from each human genome – this may not be possible if done with the human eye.”

Xiaoxu Yang, postdoctoral researcher at Dr. Gleeson’s Laboratory of Pediatric Brain Disease

This sort of computer-generated knowledge is generally known as convolutional neural network-based deep learning and has been around for the reason that Seventies. Back then, neural networks were already being built to mimic human visual processing. It has just taken just a few many years for researchers to develop accurate, efficient systems for this kind of modeling.

“The goal of machine learning and deep learning is commonly to coach the computers for prediction or classification tasks on labeled data. When the trained models are proven to be accurate and efficient, researchers would use the learned information – quite than manual annotation to process large amounts of knowledge,” explained Xin Xu, a former undergraduate research assistant in Gleeson’s lab and now an information scientist at Novartis. “We’ve come a good distance over the past 40 years in developing machine learning and deep learning algorithms, but we’re still using that very same concept that replicates the human’s ability to process data.”

Xu is referring to the knowledge needed for higher understanding diseases caused when abnormal mosaics overtake normal cells. Yang and Xu work in a laboratory that goals to do exactly that – higher understand these mosaics that result in diseases – similar to epilepsy, congenital brain disorders and more.

“Deep learning approaches are lots more efficient and their ability to detect hidden structures and connections throughout the data sometimes even surpass human ability,” Xu said. “We will process data a lot faster in this fashion, which leads us more quickly to needed knowledge.”

Source:

University of California San Diego

Journal references: