Canadian Institute for Advanced Research

Toronto, Canada

Canadian Institute for Advanced Research

Toronto, Canada
SEARCH FILTERS
Time filter
Source Type

News Article | May 2, 2017
Site: www.eurekalert.org

Donnelly Centre researchers have developed a deep learning algorithm that can track proteins, to help reveal what makes cells healthy and what goes wrong in disease Donnelly Centre researchers have developed a deep learning algorithm that can track proteins, to help reveal what makes cells healthy and what goes wrong in disease. "We can learn so much by looking at images of cells: how does the protein look under normal conditions and do they look different in cells that carry genetic mutations or when we expose cells to drugs or other chemical reagents? People have tried to manually assess what's going on with their data but that takes a lot of time," says Benjamin Grys, a graduate student in molecular genetics and a co-author on the study. Dubbed DeepLoc, the algorithm can recognize patterns in the cell made by proteins better and much faster than the human eye or previous computer vision-based approaches. In the cover story of the latest issue of Molecular Systems Biology , teams led by Professors Brenda Andrews and Charles Boone of the Donnelly Centre and the Department of Molecular Genetics, also describe DeepLoc's ability to process images from other labs, illustrating its potential for wider use. From self-driving cars to computers that can diagnose cancer, artificial intelligence (AI) is shaping the world in ways that are hard to predict, but for cell biologists, the change could not come soon enough. Thanks to new and fully automated microscopes, scientists can collect reams of data faster than they can analyze it. "Right now, it only takes days to weeks to acquire images of cells and months to years to analyze them. Deep learning will ultimately bring the timescale of this analysis down to the same timescale as the experiments," says Oren Kraus, a lead co-author on the paper and a graduate student co-supervised by Andrews and Professor Brendan Frey of the Donnelly Centre and the Department of Electrical and Computer Engineering. Andrews, Boone and Frey are also Senior Fellows at the Canadian Institute for Advanced Research. Similar to other types of AI, in which computers learn to recognize patterns in data, DeepLoc was trained to recognize diverse shapes made by glowing proteins -- labeled a fluorescent tag that makes them visible--in cells. But unlike computer vision that requires detailed instructions, DeepLoc learns directly from image pixel data, making it more accurate and faster. Grys and Kraus trained DeepLoc on the teams' previously published data that shows an area in the cell occupied by more than 4,000 yeast proteins--three quarters of all proteins in yeast. This dataset remains the most complete map showing exact position for a vast majority of proteins in any cell. When it was first released in 2015, the analysis was done with a complex computer vision and machine learning pipeline that took months to complete. DeepLoc crunched the data in a matter of hours. DeepLoc was able to spot subtle differences between similar images. The initial analysis identified 15 different classes of proteins, each representing distinct neighbourhoods in the cell; DeepLoc identified 22 classes. It was also able to sort cells whose shape changed due to a hormone treatment, a task that the previous pipeline couldn't complete. Grys and Kraus were able to quickly retrain DeepLoc with images that differed from the original training set, showing that it can be used to process data from other labs. They hope that others in the field, where looking at images by eye is still the norm, will adopt their method. "Someone with some coding experience could implement our method. All they would have to do is feed in the image training set that we've provided and supplement this with their own data. It takes only an hour or less to retrain DeepLoc and then begin your analysis," says Grys. In addition to sharing DeepLoc with the research community, Kraus is working with Jimmy Ba to commercialize the method through a new start-up, Phenomic AI. Ba is a graduate student of AI pioneer Geoffrey Hinton, a retired U of T professor and Chief Scientific Adviser of the newly established Vector Institute. Their goal is to analyse cell image-based data for pharmaceutical companies. "In an image based drug screen, you can actually figure out how the drugs are affecting different cells based on how they look rather than some simplified parameters such as live/dead or cell size. This way you can extract a lot more information about cell state form these screens. We hope to make the early drug discovery process all the more accurate by finding more subtle effects of chemical compounds," says Kraus.


News Article | May 5, 2017
Site: www.scientificamerican.com

Adapted from Born Anxious: The Lifelong Impact of Early Life Adversity—and How to Break the Cycle, by Daniel P. Keating. Published by St. Martin’s Press. Copyright © 2017 Daniel P. Keating. Reprinted with permission of the publisher, St. Martin’s Press. All rights reserved. By the late 1990s, our group at the Canadian Institute for Advanced Research had identified robust connections between early adversity and lifelong anxiety and stress, leading to problems in social relationships and mental and physical health—and even to shorter lives. What we needed was an explanation for this was happening: How does early-life stress “get under the skin”? Enter Michael Meaney, a professor at McGill University who specialized in neurology, stress, maternal care, and gene expression. He had been studying rodents displaying stress dysregulation (SDR), who were over-reactive to stressors and stayed in a stressed-out state longer. He had discovered physiological differences and behavioral problems in rats who’d been deprived of maternal nurturing, which aligned with previous studies, but he also arrived with a brand-new and as yet unpublished finding. He had actually found a biological mechanism—a process that seemed to explain why those who experienced stress early in life had so much trouble thereafter. As he explained what he had learned, we suddenly realized that this was the missing piece of our puzzle. Meaney’s lab had been studying the link from deficiencies in early nurturing to SDR for some time and had been seeking the underlying biology of why this happened. A chance meeting at a conference with a McGill colleague, Moshe Szyf, provided the inspiration. A pioneer in the growing field of epigenetics, Szyf suggested that an epigenetic change to genes that control the stress system might be worth exploring. Up to this point, nearly all the work on epigenetics had looked at it in terms of normal fetal development—where it plays a major role in controlling how and when genes work—or in response to physical inputs throughout life. The spark here was to explore whether social experiences—in this case, early nurturing—could have a similar effect. Ordinarily, our stress response system amps up or powers down proportionate to threats we face. If there’s a lion about to pounce, or a man with a gun walking our way, the system releases cortisol, which puts us on high alert. When the threat passes, the cortisol is shut off. Well, it turned out that when Michael’s newborn rats experienced the stress of poor or missing maternal nurturance at a high enough level, something happened that prevented the cortisol from being shut off. This process is what’s known as an epigenetic change: a gene’s function is altered—either switched on or switched off—by an external factor. In this case, the external factor was extreme childhood stress without comfort, and it caused an epigenetic change called “stress methylation.” Methylation means that a methyl group—a specific type of chemical molecule—has attached itself to the on-off switch that is a part of every gene. In the particular case of stress methylation, the gene whose job it is to tell the HPA axis, which controls the body's response to stress, to stand down—to shut off the flow of cortisol—is silenced. High levels of stress experienced in early life can methylate the key gene that controls this stress system. When this happens, we live as if constantly facing the pouncing lion or the man with the gun. There it is, I thought, there’s our answer: stress can get under the skin, changing the very way our genes function. I was far from alone in recognizing how this changed the landscape of how to look at early life stress. As my colleague, Clyde Hertzman, who had a penchant for pithy and pointed conclusions, remarked on first hearing about this new social epigenetic story of the biological embedding of SDR in early life: “It’s a way of getting a message to a newborn that it’s a dangerous world out there, so you’d better live hard, live fast, and, very probably, die young.” The minute we learned of this epigenetic effect we realized the potential implication for understanding not just people at the lower end of social status, but all of us. Clearly, it fit with the social inequality story we had been pursuing for a decade. Low socioeconomic status (SES) as a marker of early life adversity, with the lifelong consequences we had come to understand, was a natural fit for this new story. But it went well beyond that. Difficulties in early nurturing arise from many other sources than economic and social disadvantage. In the modern world, the stresses of managing dual careers or the worries about the hypercompetitive world that one’s children may face can interfere with the kind of nurturing that infants need. At a later point, we learned that this epigenetic change could follow another social pathway: If the mother is hyper-stressed during pregnancy, the same stress methylation can follow. Parents from lower SES groups may have a greater risk of stressful pregnancies or stressed-out early parenting, but it can happen at any level of SES—which was entirely consistent with our findings on how stress can show up at any level of society: it can happen to anyone. Our team was not alone in grasping the profound implications of this dramatic new science; soon researchers around the world joined this exploration, revolutionizing the way we look at child development. We can say with certainty that stress can change the way our genes work, with consequences across the lifespan. And beyond. It turns out that this epigenetic change—which doesn’t affect the DNA at all—can be passed down to the next generation; this has been found in animal studies, but there is recent evidence that it happens to us, too. This remarkable finding means that the social experience of early adversity can make a change that becomes part of our biology—and part of our biological inheritance: nurture becoming nature.


News Article | May 2, 2017
Site: phys.org

"We can learn so much by looking at images of cells: how does the protein look under normal conditions and do they look different in cells that carry genetic mutations or when we expose cells to drugs or other chemical reagents? People have tried to manually assess what's going on with their data but that takes a lot of time," says Benjamin Grys, a graduate student in molecular genetics and a co-author on the study. Dubbed DeepLoc, the algorithm can recognize patterns in the cell made by proteins better and much faster than the human eye or previous computer vision-based approaches. In the cover story of the latest issue of Molecular Systems Biology , teams led by Professors Brenda Andrews and Charles Boone of the Donnelly Centre and the Department of Molecular Genetics, also describe DeepLoc's ability to process images from other labs, illustrating its potential for wider use. From self-driving cars to computers that can diagnose cancer, artificial intelligence (AI) is shaping the world in ways that are hard to predict, but for cell biologists, the change could not come soon enough. Thanks to new and fully automated microscopes, scientists can collect reams of data faster than they can analyze it. "Right now, it only takes days to weeks to acquire images of cells and months to years to analyze them. Deep learning will ultimately bring the timescale of this analysis down to the same timescale as the experiments," says Oren Kraus, a lead co-author on the paper and a graduate student co-supervised by Andrews and Professor Brendan Frey of the Donnelly Centre and the Department of Electrical and Computer Engineering. Andrews, Boone and Frey are also Senior Fellows at the Canadian Institute for Advanced Research. Similar to other types of AI, in which computers learn to recognize patterns in data, DeepLoc was trained to recognize diverse shapes made by glowing proteins—labeled a fluorescent tag that makes them visible—in cells. But unlike computer vision that requires detailed instructions, DeepLoc learns directly from image pixel data, making it more accurate and faster. Grys and Kraus trained DeepLoc on the teams' previously published data that shows an area in the cell occupied by more than 4,000 yeast proteins—three quarters of all proteins in yeast. This dataset remains the most complete map showing exact position for a vast majority of proteins in any cell. When it was first released in 2015, the analysis was done with a complex computer vision and machine learning pipeline that took months to complete. DeepLoc crunched the data in a matter of hours. DeepLoc was able to spot subtle differences between similar images. The initial analysis identified 15 different classes of proteins, each representing distinct neighbourhoods in the cell; DeepLoc identified 22 classes. It was also able to sort cells whose shape changed due to a hormone treatment, a task that the previous pipeline couldn't complete. Grys and Kraus were able to quickly retrain DeepLoc with images that differed from the original training set, showing that it can be used to process data from other labs. They hope that others in the field, where looking at images by eye is still the norm, will adopt their method. "Someone with some coding experience could implement our method. All they would have to do is feed in the image training set that we've provided and supplement this with their own data. It takes only an hour or less to retrain DeepLoc and then begin your analysis," says Grys. In addition to sharing DeepLoc with the research community, Kraus is working with Jimmy Ba to commercialize the method through a new start-up, Phenomic AI. Ba is a graduate student of AI pioneer Geoffrey Hinton, a retired U of T professor and Chief Scientific Adviser of the newly established Vector Institute. Their goal is to analyse cell image-based data for pharmaceutical companies. "In an image based drug screen, you can actually figure out how the drugs are affecting different cells based on how they look rather than some simplified parameters such as live/dead or cell size. This way you can extract a lot more information about cell state form these screens. We hope to make the early drug discovery process all the more accurate by finding more subtle effects of chemical compounds," says Kraus. Explore further: New map uncovers the traffic of life in a cell More information: Oren Z Kraus et al. Automated analysis of high‐content microscopy data with deep learning, Molecular Systems Biology (2017). DOI: 10.15252/msb.20177551


News Article | May 3, 2017
Site: www.eurekalert.org

More use of technology led to increases in attention, behavior and self-regulation problems for adolescents already at risk for mental health issues, new study finds DURHAM, N.C. -- More use of technology is linked to later increases in attention, behavior and self-regulation problems for adolescents already at risk for mental health issues, a new study from Duke University finds. "Also, on days at-risk adolescents use technology more, they experience more conduct problems and higher ADHD symptoms compared to days they use technology less," said Madeleine J. George, a Duke Ph.D. candidate and the lead author of the study. However, the study also found that using technology was linked to some positive outcomes: On days when adolescents spent more time using digital technologies they were less likely to report symptoms of depression and anxiety. The research, published May 3 in a special issue of Child Development, looks at associations between adolescents' mental health symptoms and how much time they spent each day texting, using social media and using the Internet. For the study, 151 young adolescents completed surveys on smartphones about their daily digital technology use. They were surveyed three times a day for a month and were assessed for mental health symptoms 18 months later. The youth participating were between 11 and 15 years old, were of a lower socioeconomic status and were at a heightened risk for mental health problems. The adolescents spent an average of 2.3 hours a day using digital technologies. More than an hour of that time was spent texting, with the adolescents sending an average of 41 texts a day. The researchers found that on days when adolescents used their devices more -- both when they exceeded their own normal use and when they exceeded average use by their peers -- they were more likely to experience conduct problems such as lying, fighting and other behavioral problems. In addition, on days when adolescents used digital devices more, they had difficulty paying attention and exhibited attention deficit-hyperactivity disorder symptoms. The study also found that young adolescents who spent more time online experienced increases in conduct problems and problems with self-regulation -- the ability to control one's behavior and emotions -- 18 months later. It's unclear whether high levels of technology use were simply a marker of elevated same-day mental health symptoms or if the use of technology exacerbated existing symptoms, said Candice Odgers, the senior author of the study and a professor in Duke's Sanford School of Public Policy. On the positive side, the researchers found evidence that digital technology use may be helpful to adolescents experiencing depression and anxiety. More time spent texting was associated with fewer same-day symptoms of depression and anxiety. "This finding makes sense when you think about how kids are commonly using devices to connect with their peers and social networks," said Odgers, a faculty fellow at the Duke Center for Child and Family Policy. The findings suggest contemporary youth may be using digital technology to connect in positive ways versus isolating themselves, the authors said. In the past, some research found that teenagers using digital technology were socially isolated. But at that time, only a small minority of youth were frequently online. Odgers noted that the adolescents in the study were already at an increased risk for mental health problems regardless of digital device use. It's therefore unclear if the findings would apply to all adolescents. Because this was a correlational study, it is possible factors other than technology use could have caused the increase in mental health problems. As rates of adolescent technology use continue to climb, more work is needed to investigate its effects, the researchers say. Odgers and George are now conducting a large study of more than 2,000 N.C. adolescents to determine how and why high digital device use predicts future problems among some adolescents. The study also looks at whether being constantly connected during adolescence could provide opportunities to improve mental health. This study was supported by the William T. Grant Foundation and the Verizon Foundation. Russell was supported by the National Institute on Drug Abuse (T32 DA017629, P50 DA010075 and P50 DA039838). Odgers is a Jacobs Foundation Advanced Research Fellow and a fellow of the Canadian Institute for Advanced Research. CITATION: "Concurrent and Subsequent Associations between Daily Digital Technology Use and High-Risk Adolescents' Mental Health Symptoms," Madeleine J. George, Michael A. Russell, Joy R. Piontak and Candice L. Odgers. Child Development, May 3, 2017. DOI: 10.1111/cdev.12819


Wu W.,Université de Sherbrooke | Tremblay A.-M.-S.,Université de Sherbrooke | Tremblay A.-M.-S.,Canadian Institute for Advanced Research
Physical Review X | Year: 2015

Superconductivity in heavy-fermion materials can sometimes appear in the incoherent regime and in proximity to an antiferromagnetic quantum critical point. Here, we study these phenomena using largescale determinant quantum Monte Carlo simulations and the dynamical cluster approximation with various impurity solvers for the periodic Anderson model with frustrated hybridization. We obtain solid evidence for a dx2-y2 superconducting phase arising from an incoherent normal state in the vicinity of an antiferromagnetic quantum critical point. There is a coexistence region, and the width of the superconducting dome increases with frustration. Through a study of the pairing dynamics, we find that the retarded spin fluctuations give the main contribution to the pairing glue. These results are relevant for unconventional superconductivity in the Ce-115 family of heavy fermions.


Chudek M.,University of British Columbia | Henrich J.,University of British Columbia | Henrich J.,Canadian Institute for Advanced Research
Trends in Cognitive Sciences | Year: 2011

Diverse lines of theoretical and empirical research are converging on the notion that human evolution has been substantially influenced by the interaction of our cultural and genetic inheritance systems. The application of this culture-gene coevolutionary approach to understanding human social psychology has generated novel insights into the cognitive and affective foundations of large-scale cooperation, social norms and ethnicity. This approach hypothesizes a norm-psychology: a suite of psychological adaptations for inferring, encoding in memory, adhering to, enforcing and redressing violations of the shared behavioral standards of one's community. After reviewing the substantial body of formal theory underpinning these predictions, we outline how this account organizes diverse empirical findings in the cognitive sciences and related disciplines. Norm-psychology offers explanatory traction on the evolved psychological mechanisms that underlie cultural evolution, cross-cultural differences and the emergence of norms. © 2011 Elsevier Ltd.


Grant
Agency: GTR | Branch: EPSRC | Program: | Phase: Training Grant | Award Amount: 3.99M | Year: 2014

The Scottish Doctoral Training Centre in Condensed Matter Physics, known as the CM-DTC, is an EPSRC-funded Centre for Doctoral Training (CDT) addressing the broad field of Condensed Matter Physics (CMP). CMP is a core discipline that underpins many other areas of science, and is one of the Priority Areas for this CDT call. Renewal funding for the CM-DTC will allow five more annual cohorts of PhD students to be recruited, trained and released onto the market. They will be highly educated professionals with a knowledge of the field, in depth and in breadth, that will equip them for future leadership in a variety of academic and industrial careers. Condensed Matter Physics research impacts on many other fields of science including engineering, biophysics, photonics, chemistry, and materials science. It is a significant engine for innovation and drives new technologies. Recent examples include the use of liquid crystals for displays including flat-screen and 3D television, and the use of solid-state or polymeric LEDs for power-saving high-illumination lighting systems. Future examples may involve harnessing the potential of graphene (the worlds thinnest and strongest sheet-like material), or the creation of exotic low-temperature materials whose properties may enable the design of radically new types of (quantum) computer with which to solve some of the hardest problems of mathematics. The UKs continued ability to deliver transformative technologies of this character requires highly trained CMP researchers such as those the Centre will produce. The proposed training approach is built on a strong framework of taught lecture courses, with core components and a wide choice of electives. This spans the first two years so that PhD research begins alongside the coursework from the outset. It is complemented by hands-on training in areas such as computer-intensive physics and instrument building (including workshop skills and 3D printing). Some lecture courses are delivered in residential schools but most are videoconferenced live, using the well-established infrastructure of SUPA (the Scottish Universities Physics Alliance). Students meet face to face frequently, often for more than one day, at cohort-building events that emphasise teamwork in science, outreach, transferable skills and careers training. National demand for our graduates is demonstrated by the large number of companies and organisations who have chosen to be formally affiliated with our CDT as Industrial Associates. The range of sectors spanned by these Associates is notable. Some, such as e2v and Oxford Instruments, are scientific consultancies and manufacturers of scientific equipment, whom one would expect to be among our core stakeholders. Less obviously, the list also represents scientific publishers, software houses, companies small and large from the energy sector, large multinationals such as Solvay-Rhodia and Siemens, and finance and patent law firms. This demonstrates a key attraction of our graduates: their high levels of core skills, and a hands-on approach to problem solving. These impart a discipline-hopping ability which more focussed training for specific sectors can complement, but not replace. This breadth is prized by employers in a fast-changing environment where years of vocational training can sometimes be undermined very rapidly by unexpected innovation in an apparently unrelated sector. As the UK builds its technological future by funding new CDTs across a range of priority areas, it is vital to include some that focus on core discipline skills, specifically Condensed Matter Physics, rather than the interdisciplinary or semi-vocational training that features in many other CDTs. As well as complementing those important activities today, our highly trained PhD graduates will be equipped to lay the foundations for the research fields (and perhaps some of the industrial sectors) of tomorrow.


Gingras M.J.P.,University of Waterloo | Gingras M.J.P.,Perimeter Institute for Theoretical Physics | Gingras M.J.P.,Canadian Institute for Advanced Research | McClarty P.A.,Max Planck Institute for the Physics of Complex Systems
Reports on Progress in Physics | Year: 2014

The spin ice materials, including Ho2Ti2O7 and Dy 2Ti2O7, are rare-earth pyrochlore magnets which, at low temperatures, enter a constrained paramagnetic state with an emergent gauge freedom. Spin ices provide one of very few experimentally realized examples of fractionalization because their elementary excitations can be regarded as magnetic monopoles and, over some temperature range, spin ice materials are best described as liquids of these emergent charges. In the presence of quantum fluctuations, one can obtain, in principle, a quantum spin liquid descended from the classical spin ice state characterized by emergent photon-like excitations. Whereas in classical spin ices the excitations are akin to electrostatic charges with a mutual Coulomb interaction, in the quantum spin liquid these charges interact through a dynamic and emergent electromagnetic field. In this review, we describe the latest developments in the study of such a quantum spin ice, focusing on the spin liquid phenomenology and the kinds of materials where such a phase might be found. © 2014 IOP Publishing Ltd.


Gardner J.S.,U.S. National Institute of Standards and Technology | Gardner J.S.,Indiana University | Gingras M.J.P.,University of Waterloo | Gingras M.J.P.,Canadian Institute for Advanced Research | Greedan J.E.,McMaster University
Reviews of Modern Physics | Year: 2010

Within the past 20 years or so, there has occurred an explosion of interest in the magnetic behavior of pyrochlore oxides of the type A2 3+ B2 4+ O7, where A is a rare-earth ion and B is usually a transition metal. Both the A and B sites form a network of corner-sharing tetrahedra which is the quintessential framework for a geometrically frustrated magnet. In these systems the natural tendency to form long-range ordered ground states in accord with the third law of thermodynamics is frustrated, resulting in some novel short-range ordered alternatives, such as spin glasses, spin ices, and spin liquids, and much new physics. This article attempts to review the myriad of properties found in pyrochlore oxides, mainly from a materials perspective, but with an appropriate theoretical context. © 2010 The American Physical Society.


Maciejko J.,University of Alberta | Maciejko J.,Canadian Institute for Advanced Research | Fiete G.A.,University of Texas at Austin
Nature Physics | Year: 2015

Topological insulators have emerged as a major topic of condensed matter physics research, with several novel applications proposed. Although there are now a number of established experimental examples of materials in this class, all of them can be described by theories based on electronic band structure, which implies that they do not possess electronic correlations strong enough to fundamentally change this theoretical description. Here, we review recent theoretical progress in the description of a class of strongly correlated topological insulators - fractionalized topological insulators - where band theory fails owing to the fractionalization of the electron into other degrees of freedom.

Loading Canadian Institute for Advanced Research collaborators
Loading Canadian Institute for Advanced Research collaborators