Advanced research computing key to boosting impact across all disciplines
Compute Canada supported research ranks above both world and Canadian citation averages across all disciplines, says new study
Dr. George Wells was part of a group of scientists who published research in 2012 identifying 15 newly-discovered genetic variations that boost the risk of coronary artery disease. The largest genetic study of its kind at the time, it compared nearly 64,000 people with the disease to about 131,000 people without it.
The research was significant for two reasons: it provided insights that may eventually help us better understand and treat the most common cause of death, and it was done with the help of advanced research computing (ARC). In fact, the study would not have been possible without ARC.
Dr. Wells is director of the Cardiovascular Research Methods Centre at the University of Ottawa Heart Institute and professor in the School of Epidemiology, Public Health and Preventive Medicine at the University of Ottawa. He is also just one of a growing number of researchers who are increasing their output and influence by using some of the world’s most advanced digital tools.
In June 2016, Compute Canada released the first Canadian study to measure in a quantitative way the research output enabled by advanced research computing. In a nutshell, it found that ARC is a key ingredient for boosting impact across all disciplines and at all institutions.
The study identified more than 70,000 scientific outputs (including journal and conference publications) produced since 2010 by reviewing the CVs of more than 2,300 researchers who are active Compute Canada users. Of those outputs, nearly half had been enabled by Compute Canada. The disciplines most likely to use digital tools were physics, environmental and Earth science, followed by math and statistics, and engineering.
Researchers in personalized medicine and genetics were significant users. As Dr. Wells did, Dr. Mark Lathrop and his team used ARC to analyze large amounts of genetic information linked to disease. Dr. Lathrop is a professor in the Department of Human Genetics at McGill University and scientific director of the McGill University and Génome Québec Innovation Centre. In his case, it was about identifying 11 genomic regions newly associated with Alzheimer’s disease, which could eventually lead to new treatments.
Likewise, Dr. Guy Rouleau used ARC to estimate the degree to which genetic variations are shared among five psychiatric disorders. Dr. Rouleau is director of the Montreal Neurological Institute – McGill University. He and his team found the genetic correlation was high between schizophrenia and bipolar disorder, and moderate between schizophrenia and major depressive disorder, bipolar disorder and major depressive disorder, and attention-deficit/hyperactivity disorder and major depressive disorder. The correlation was low between schizophrenia and autism spectrum disorder.
This is important because it shows genetic risk factors are shared across key psychiatric disorders, providing pointers for future research into treatments.
“Genetics and genomics are generating huge quantities of data,” says Dr. Rouleau. “Coupled with huge phenotypic datasets, ARC is the only way to extract the true links and associations, which will drive the future of medicine.”
To determine the impact of advanced research computing resources and expert support the Compute Canada report used a metric known as Field-Weighted Citation Impact (FWCI), which is the ratio of the total citations received divided by the average for the subject field. A score of 1.6, for example, means that paper received 60 per cent more citations than the world average for that discipline. The article Dr. Wells’ group wrote on coronary artery disease weighed in with 38.6. The papers Dr. Lathrop and Dr. Rouleau and their teams wrote scored 38.2 and 40.6, respectively. In other words, all three papers received close to 40 times the average in their respective disciplines.
More generally, the Compute Canada study found that in all disciplines, the average FWCI for publications that made use of its computing powers was substantially above the world average. In some cases, it was more than double. While the Canadian FWCI average tends to be higher than the world average, Compute Canada enabled publications also consistently exceeded this higher domestic score.
“The basic message is that Compute Canada enabled papers have significantly more citations and are in higher impact journals,” says Compute Canada’s chief science officer, Dr. Dugan O’Neil, who helped compile the report.
Top research institutions also saw big gains, he says, with Compute Canada enabled publications scoring substantially higher than the institutional average. At the University of Toronto for example, the overall FWCI of its publications was just under two, while those enabled by Compute Canada scored a three. McGill University and the University of British Columbia had similar impact numbers, while York, Carleton and Simon Fraser universities saw some of the most dramatic increases in FWCI scores for Compute Canada-enabled publications.
The reality is that some research — climate modelling, artificial intelligence, subatomic physics and astronomy for example — simply can’t be done without ARC, says Dr. O’Neil. But other disciplines are also now beginning to take fuller advantage. In fact, the strongest differences in impact can be seen in areas in which ARC adoption is in relatively early stages, particularly in the social sciences and humanities.
Most dramatic rise in impact seen in humanities and social sciences
Dr. John Simpson, Compute Canada’s humanities and social sciences specialist, says researchers using texts, including historical and social media documents, are seeing particularly large gains in productivity and reach. That’s because digital tools allow researchers to make connections among pieces of text quickly and effectively. “You need advanced computing to sift through and crunch it,” says Dr. Simpson.
Dr. O’Neil agrees: “As technology has advanced, the necessity of having this infrastructure has spread from discipline to discipline. It’s only a matter of time before all research will have a digital component,” he says.
It hasn’t escaped scholars’ notice that colleagues using these technologies are getting further ahead. Dr. O’Neil, for one, sees the unrealized potential and is keen to grow Compute Canada’s ARC systems, storage and software solutions.
“We have a national platform with merit-based access but we aren’t able to meet everyone’s needs,” he says.
Yet for those researchers who are able to integrate ARC into their work, it is a recipe for delivering on significant promises. University of Waterloo digital historian Dr. Ian Milligan is one of those researchers using ARC to solve a unique problem. For Dr. Milligan, it is the limited life span of websites — their content is deleted or replaced roughly every 100 days.
“New political leader comes in, website comes down,” explains Dr. Milligan. “A social movement happens and someone doesn’t pay their server fee — website comes down.”
This means that researchers trying to reconstruct, analyze and learn from the recent past face both a challenge and an opportunity. Mining old websites can yield a wealth of data about political, cultural, social and economic trends. But first you need large-scale storage to archive sites before they disappear, and then you need computational muscle to analyse vast amounts of data.
“You can’t do historical research from the 1990s and later if you don’t use websites,” says Dr. Milligan. “We’ve lived online now for about 20 years and we need to consider what people said on the web.”
With help from Compute Canada, Dr. Milligan has done just that. He developed a database called webarchives.ca, a searchable index of web pages of 50 Canadian political parties and interest groups dating back to 2005. It makes use of data in ways that allow present and future humanities and social science scholars to access, interpret and curate vast swaths of digital-borne sources detailing the recent past.
Webarchives.ca represents just one of a growing number of projects in the humanities and social sciences harnessing the power of ARC to boost its reach and impact. Compute Canada’s analysis of Field-Weighted Citation Impact (FWCI) — the ratio of total citations divided by the average for the subject field — found the FWCI of Canadian publications enabled by Compute Canada in the social sciences and humanities was just about triple the world average.
Digital historians are paying attention: roughly 4,000 users checked out webarchives.ca within days of the site going live.
University of Guelph digital literary historian Dr. Susan Brown has made use of Compute Canada’s services to launch the Canadian Writing Research Collaboratory, an online platform designed to give scholars unprecedented avenues for studying Canada’s literary and cultural heritage.
The collaboratory is a virtual research environment that allows scholars to conduct, store, share and publish research about our country’s literary and cultural history.
Yet before launching such projects, she says researchers need to understand the breadth of the digital tools at their disposal. Humanities and social science scholars often don’t have the kind of training necessary to take full advantage.
“But Compute Canada has really stepped up in terms of training,” she says. “They understand the kind of computing humanities and social science scholars need. They know it is about big discoveries, but also about seeing things differently or asking new questions.”