Steve Wozniak Full Interview
Future of Education
Demand To Know What’s In The Products We Buy
Conversations with Experts, Past & Present: Transformation Pt. 1 with Agnis Stibe
Interview with Former US Deputy Secretary of Energy Bill Martin
Making the Most Out of Remote Work Using Technology
Future of Housing
An Interview With Aman Gupta Founder Of Fan Expo
Interview with Chris Wellise: Chief Sustainability Officer at Hewlett Packard Enterprise
“Designing successful drugs is like solving a puzzle,” the researchers explain. “Without knowing the three-dimensional shape of a protein, it would be like trying to solve that puzzle with a blindfold on.”
By: Adi Gaskell, Contributor
The outbreak of the COVID-19 virus across the world has placed the spotlight on the ability of society to swiftly develop new treatments. As the disease spread from Wuhan rapidly across the world, the efforts to devise a vaccine seemed to take an interminable length of time.
Of course, that’s grossly unfair, not least as new drugs typically require around 12-14 years to reach the market, with data suggesting this results in costs of several billion dollars. There have been various attempts to render this process more efficient, and therefore both faster and cheaper. Many of these took a degree of inspiration from a Cornell study, which suggested that better use of automation could reduce the cost of drug discovery by an impressive 70%.
Most of the use cases for automation in drug discovery today revolve around proposing avenues for analysis, but there is nonetheless some variety on show. For instance, a second study, also from Cornell, highlighted the possibilities of using a data-driven approach to try and identify potentially toxic side effects of a treatment that would prevent it from being tested on humans. The aim was to cut down on the wasted time and money of going to the human stage of clinical trials when the treatment was unlikely to get past that stage.
AI is largely being used early in the process, however, as exemplified by a project from the University of Toronto, which uses machine learning to generate 3D images of protein molecules to help the drug development process.
The idea is that by determining the atomic structure of a protein molecule, that will aid an understanding of how the protein works, and importantly, whether it may respond to drug therapies, which typically work by binding themselves to protein molecules, and then changing the shape and functioning of the molecule. The researchers explain that drugs need to bind only to those proteins needed to fight the disease, ignoring those not involved and potentially causing unwanted side effects.
Their algorithm takes microscopic images of the protein using electron cryomicroscopy, and then reconstructs them. The team believe their approach opens the door for many more proteins to be assessed than would have been possible in the past, thus greatly expanding the scope for new treatments.
Ultimately, this kind of approach works to significantly speed up the drug development process by allowing faster targeting of proteins. The team explains that whereas existing techniques can take weeks to develop 3D protein structures digitally, they are able to achieve this in a matter of minutes. What’s more, it’s possible using a single computer rather than a cluster of highly-powered machines.
“We hope this will allow discoveries to happen at a ground-breaking pace in structural biology,” the researchers conclude. “The ultimate goal is that it will directly lead to new drug candidates for diseases, and a much deeper understanding of how life works at the atomic level.”
The Canadian team has put this expertise into a startup, called Deep Genomics, which utilizes deep learning to examine genetic data and try to identify specific genes that are responsible for various diseases. These genes can then be targeted by specific treatments. To date, the company have largely focused on explore the human genome for mutations, which typically play a crucial role in disease.
British startup Benevolent AI are another company that is making strides in this field, with their approach revolving around using AI to trawl through the scientific literature for potential new avenues to explore. The company has already made significant breakthroughs in areas such as Alzheimer’s.
Spanish startup Mind the Byte is another fascinating entry into the market. They offer pharmaceutical clients an in silico drug development analysis via their SaaS platform. The ultimate aim is to make drug development more accessible to smaller biotech companies. Fellow Spanish startup Naru looks to mine the treasure trove of our genetic and lifestyle data to add valuable insights to clinical trials.
It’s clear that drug development is ripe for optimization, and while we are still in a very early stage of development to do just that, there are a number of promising signs that real progress is being made. It might be a while until we can develop the kind of personalized medicines to respond in real-time to diseases such as the Coronavirus, but the signs are certainly promising.
How Covid Might Change How We Live
Zero Energy Communities
Where Now For Smart Cities?
Sign up to our newsletter for updates on live events in your area, new content and interviews