These are the companies at the intersection of AI and biology
A folded protein. (DeepMind)
This post is sponsored by Brilliant. Sign up today and get 20% off when you use this link.
Artificial intelligence companies are known for taking on projects in a wide variety industries in hopes that one, or several, will lead to a breakthrough. Google DeepMind is no different. The company, which most famously trained an AI to beat the world champion of the board game Go, claims to have also published research that makes "significant progress on one of the core challenges in biology."
That challenge is known as protein folding, and it has long baffled scientists. One researcher, Cyrus Levinthal, estimated that it would take the life of the universe to predict the way any given protein folds. His claim, known as Levinthal's Paradox, goes like this: it would take you the life of the universe to guess how a protein will fold, but a protein itself folds spontaneously.
According to DeepMind, AI can help scientists accurately guess the shape of a protein fold using neural networks, which find patterns and measure possibilities in data far faster than humans and conventional programs ever could. Being able to estimate protein folding will help researchers understand the causes behind Parkinson's disease, Alzheimer's, and cystic fibrosis. It will also help them better understand the complex biology of the human body, and potentially create cures to our worst diseases.
Protein folding is just one area of focus, however, and DeepMind is just one startup. This post will cover three other efforts and organizations working at the intersection of AI and computational biology.
Neural networks and deep learning allow scientists to map and analyze vast sums of data, making them useful tools in drug discovery.
Deep Genomics, for example, is a firm that uses what it calls an AI Workbench to evaluate medicine candidates and develop therapies for genetic determinants of disease.
"Drug development has traditionally been a serendipitous activity, like throwing a stick into a tree and seeing if an apple falls,” says Brendan Frey, Deep Genomics founder. “This worked in the early days, but the low-hanging fruit is gone and this traditional approach is leading to more failures, greater delays and rising costs."
Deep Genomics built Project Saturn, which evaluates RNA and DNA molecules against targets with the hopes of creating new drugs that can tackle cancer, Alzheimer's, muscular disorders, and other diseases.
Cyclica is another Toronto-based company that works in drug development. It also maps the unintended proteins that drugs interact with, which is a leading cause of side effects in treatment.
"Once a drug enters the body, it interacts with dozens, if not hundreds, of proteins before it is eliminated from the body," Cyclica's web site reads. "These off-target interactions can impact the safety of a drug or may lead to drug repurposing opportunities."
"Our goal is to examine all possible proteins in the body that a drug can bind to," CEO Naheed Kurji told the National Post.
Biorelate, like others in this space, is a research company that uses natural language processing to pull insights and summarize data from scientific and medical papers. Clients use a dashboard that "auto-curates knowledge from millions of biomedical text sources, revealing insights that might otherwise be lost."
"Global scientific output doubles every 9 years," according to Biorelate's site. "Curating and connecting data to understand how new biomedical innovations can be developed optimally is becoming an impossible challenge."
Finding secondary uses for existing drugs is a smaller area of focus that aims to save on the industry's notoriously expensive and lengthy R&D.
BioExcel Therapeutics explores drugs that are either already federally approved or in late-stage development and analyzes them for non-primary uses. Doing this cuts down on extensive research and development time, according to BioExcel, and allows them to evaluate "existing approved drugs and/or clinically evaluated product candidates together with big data and proprietary machine learning algorithms to identify new therapeutic indices."