An ‘intelligent’ approach
AI and machine-learning approaches seem to be an industry theme for Q1 2018
MCLEAN, Va.—It might not have been the Super Bowl, but another “bowl” competition was announced in the first quarter of the year: the fourth annual Data Science Bowl, launched by Booz Allen Hamilton and Kaggle and hosted in partnership with the Broad Institute of MIT and Harvard, NVIDIA and PerkinElmer. The Data Science Bowl is a 90-day competition in which participants are challenged to train deep learning models to assess images of cells and identify nuclei, without human intervention and regardless of experimental setup. Those who create the best algorithms will split $170,000 in cash and prizes, including an NVIDIA DGX Station, a personal artificial intelligence (AI) supercomputer that provides a computing capacity of 400 CPUs in a desktop workstation.
“This year’s Data Science Bowl will bring together thousands of people from around the world to confront deadly diseases in one of our most complex challenges yet,” said Ray Hensberger, a Booz Allen Hamilton principal. “Despite some progress, it remains time-consuming and expensive to find treatments for all types of diseases. We believe that pairing artificial intelligence and the collective ingenuity of the global data science community will yield powerful tools that can help accelerate the search for medical cures.”
The Broad Institute will provide participants “with data from thousands of nuclei from a wide variety of imaging experiments,” a Booz Allen Hamilton press release noted, which they are then challenged to use to create algorithms capable of identifying nuclei in any cell image. Per the release, this competition could help fill a needed void in AI-based early-stage research: “All current options for nuclei detection require time-consuming biologist intervention. There are no deep learning models available today that can identify nuclei across multiple experimental setups and testing conditions. Finally, biologists often do not have the technical expertise nor the time needed to train their own deep learning models.”
“We’re proud to bring together some of the world’s top minds in a forum that harnesses data science for social good,” remarked Anthony Goldbloom, CEO of Kaggle. “This year’s competition will provide our global community with an exciting opportunity to transform biomedical research through new open-source solutions.”
In the past three years, the 2015, 2016 and 2017 Data Science Bowls focused on applying deep learning to marine health, heart disease and lung cancer, respectively. The winners developed algorithms to automatically classify images of plankton (key indicators of the health of an ocean), to automate the analysis of cardiac MRI images and to more accurately assess which lung lesions are cancerous.
On the other side of the pond, Nestlé and Nuritas have kicked off an AI collaboration of their own, focused on the discovery of bioactive peptide networks in certain target areas. The companies will be applying Nuritas’ award-winning novel technology platform, which unites artificial intelligence with DNA analysis to predict, unlock and validate efficacious peptides from natural food sources. In turn, Nestlé will leverage its own expertise to validate any new peptides and determine their efficacy within the targeted areas of interest.
Nuritas’ approach begins with elucidating the disease and targets of choice, then using their platform “to identify the characteristics specific to our area of focus.” Then comes the application of their AI algorithms, including deep learning, “to predict which novel food-derived bioactive peptides deliver the predetermined effect” that is needed, a process that the company claims “cuts out many thousands of hours of trial and error.”
“At Nuritas our mission is to positively impact billions of lives worldwide, and we therefore are delighted to be collaborating with Nestlé, the world’s largest food and beverage company on such an important project. We are really looking forward to beginning this impactful journey together,” said Nora Khaldi, founder and chief scientific officer of Nuritas.
“As our understanding of food and nutrition continues to grow, our global research and development network is looking ahead to discover how we can help enhance quality of life and contribute to a healthier future for everyone. Research partnerships such as that with Nuritas help us achieve that goal,” added Richard Stadler of the Nestlé Research Centre.
Other discovery and validation efforts are underway in the British Isles as well. Medical research charity LifeArc (previously MRC Technology) and the Milner Therapeutics Institute at the University of Cambridge have partnered to identify and validate drug targets in immuno-oncology and respiratory diseases. The Milner Therapeutics Institute is a fully integrated institute of the School of Biological Sciences and the School of Clinical Medicine at the University of Cambridge, and for its part, LifeArc has helped contribute to the development of drugs such as Keytruda, Actemra, Tysabri and Entyvio.
“Drug discovery is a long and risky process, and our collaboration with the Milner Therapeutics Institute represents a powerful way to unlock new potential approaches to help patients,” Dr. Justin Bryans, executive director of drug discovery at LifeArc, said in a press release. “We are excited about the opportunity to work at the interface of drug discovery and AI and apply the knowledge in this field to help expedite the delivery of new treatments to patients.”
The partnership will unite the machine-learning and bioinformatics of the Milner Therapeutics Institute with LifeArc’s experience in drug discovery in hopes of not just discovering new targets, but developing new machine learning-based approaches to discovering novel therapeutic targets, stratifying patient populations and predicting drug efficacy. LifeArc’s capabilities include assay development, screening with a library of more than 120,000 compounds, hit-to-lead optimization, proof-of-concept work and ADME/DMPK.
Prof. Tony Kouzarides, director of the Milner Therapeutics Institute, commented: “We are delighted to be working closely with LifeArc in applying artificial intelligence and machine learning approaches to drug discovery. There is a lot of interest in these methods for the potential benefit of patients. The drug discovery insight and investment from LifeArc will be important in realizing this.”
In other U.K. partnering news, Intellegens has launched a commercial collaboration with e-Therapeutics, the Oxford, U.K.-based pioneer of Network-Driven Drug Discovery. Under the collaboration, e-Therapeutics will apply Intellegens’ Alchemite AI platform to produce improved predictions, fix errors and fill gaps in the large-scale biological and chemical information repositories in e-Therapeutics’ proprietary databases.
“Alchemite is the first in a series of application-specific AI modules that we are developing at Intellegens,” said Dr. Gareth Conduit, chief technology officer and co-founder of Intellegens. “These will be designed to address specific, high-value data analysis bottlenecks that we are uncovering through our discussions with existing and potential customers. With these new modules, we intend to pursue new business opportunities in both the life sciences and other sectors.”
Conduit developed Alchemite after years of research, presenting a new method for analyzing sparsely populated matrices via deep neural networks and novel machine learning approaches. This AI engine features Intellegens’ deep learning algorithm, which the company notes on its website is “capable of training models from data [which] can be as little as 0.05 percent complete. Trained models can be used to make new predictions, identify errors and maximize a set of desired parameters.”
Dr. Jonny Wray, head of discovery informatics at e-Therapeutics, noted that: “We already utilize machine learning heavily in our discovery platform to augment empirical biological and chemical data. Our partnership with Intellegens will enhance and extend our internal capabilities at the cutting edge of AI research and application.”