[ad_1]
The drug discovery had to change. The fruits at hand had been harvested, but the biopharmaceutical industry, according to the words of Deep Genomics CEO, Brendan Frey, still pushes the tree up until an apple falls .
"Doing drugs has always been a game of chance. Big Pharma throws a stick into the tree and sees what happens, "Frey told FierceBiotech. "It's like Big Pharma companies enter a casino, put a million dollar coin in a slot machine and with a probability of 10% or more, they win."
Instead of playing to get the fruit higher up the tree, Frey created Deep Genomics, a company that uses artificial intelligence to discover new disease targets and the best compounds for drug use. He calls this the construction of a ladder.
Since 2015, Toronto-based biotech has quietly silenced its AI Workbench, a technology based on – you guessed it – AI to discover genetic drugs. It uses more than 20 machine learning systems that have been "carefully validated and tested" and trained in public and proprietary data to screen for disease-causing mutations in search of new drug targets.
RELATED: Unlocking Small Data, the "Next Frontier" of Drug Discovery
"We have developed a system that, in less than two hours, can analyze more than 200,000 mutations of pathogenic patients and automatically identify potential drug targets," Frey said. Instead of chasing known targets or entering a hot therapeutic area, Deep Genomics lets the system choose – and it has landed on Wilson's disease, a rare genetic disease that does not have any disease-modifying treatment.
People with Wilson's disease can not eliminate excess copper, which ends up accumulating in various tissues such as the liver and brain. Current medications aim to reduce the level of copper by preventing the body from absorbing the copper contained in food or causing it to get rid of copper through the urine. If it is not treated, Wilson's disease is fatal.
There is no treatment to allow patients to regain the ability to remove copper, as researchers have struggled to understand how the underlying mutation had led to Wilson's disease.
"Mutations can cause diseases in different ways and humans are very familiar with some of these methods, for example when a mutation alters a protein so that it no longer works properly," Frey said. "But there is a huge area of change that is not a problem with this type of mechanism."
RELATED: GSK hires Branson of Genentech to strengthen its artificial intelligence team
And that's where the AI comes in. The Deep Genomics system discovered that the mutation was altering an amino acid from ATP7B, a copper-binding protein absent in Wilson's patients. But he also predicted that this change should not affect the functioning of the protein. Artificial intelligence eventually discovered that the mutation was upsetting an instruction in the genome telling cells how to make that protein, thus preventing it from being manufactured at all.
The platform identified a dozen potential drug candidates and Deep Genomics took them to the lab to make sure they were working as planned by the AI. After some tolerability and pharmacokinetic experiments, the company said DG12P1 was its first potential candidate in the industry, which would allow it to progress to IND.
"We see this first statement as a new era for drug discovery, for us, Deep Genomics, and more broadly," said Frey. "Because it's really the first time that artificial intelligence has really helped many steps of drug discovery."
Other companies have deployed this technology at different stages of discovery. Insilico Medicine recently released a study showing that in just three weeks, its algorithm has detected 30,000 new compounds to target the domain-specific receptor for discoidin 1, or DDR1, involved in fibrotic diseases. Atomwise's technology analyzes virtual compounds to identify new drugs that may take years to use traditional discovery methods. And Insitro uses in vitro systems to predict what drug developers will see in a human clinical model. It is these models that would propose drug targets and predict how patients, or specific subgroups of patients, would respond to certain treatments.
For Deep Genomics, it took 18 months to move from a brand new target to a drug candidate. And he can name new programs even faster.
RELATED: Stealthy Insitro opens its doors – starting with a contract with Gilead of up to $ 1.05 billion
"We expect to declare two candidates this year, we expect at least twice as much next year and at least twice as much as the next year," he said.
The challenge will be how Deep Genomics will develop all of these candidates and pass them on to patients.
"We will do this by combining their internal development and their partnership. This is something we would do very carefully because we want to make sure that the drugs are developed effectively, "said Frey.
Will the Deep Genomics approach become the biopharmaceutical scale needed to go further in the apple tree? Time will tell us.
[ad_2]
Source link