In the near future, drugs like psilocybin, MDMA and LSD may be considered passé, surpassed by the next generation of psychedelics. Some venture capitalists and researchers are banking on yet-to-be developed drugs that they feel greatly “improve” upon the so-called “classic” psychedelics in that they won’t cause hallucinations and may potentially be more effective in treating depression, addiction, and mental health.
Read: What You Should Know About Psilocybin For Depression
While research is still in the early stages, scientists are making massive advancements by deploying artificial intelligence and machine learning to sift through hundreds of millions of different drugs, analogs and derivatives of psychedelics. Usually drug companies focus on developing one drug for one condition at a time. But using advanced computer science, some researchers are instead developing dozens of potential psychedelics—or drugs that resemble them—all at once.
The medications that are designed next may look unlike anything seen in psychiatry before—at least, if these predictions can be backed up by science. The use of AI not only holds promise to push new psychedelic molecules through clinical trials faster, but it could also radicalize the way we think about mental health.
Need Access to Shrooms?
Secure your supply with DoubleBlind!
MagicMed Industries, a biotech startup based in Calgary, Alberta is building what it calls its “Psybary,” a large portfolio of patented molecules based on psychedelics. According to Green Market Report, the Psybary includes 12 filed patents related to nine different drug categories covering 125 million individual molecules.
That seems like a lot of drugs, and it is, but most of them are the smallest atomic variation on already established psychedelics like psilocybin. It’s not yet clear if the majority of these analogs would even be psychoactive, let alone effective at treating mental health. And to be clear, these patents haven’t been awarded yet. But using supercomputing and machine learning, companies like MagicMed are gambling that they can find effective new drugs in record time.
“A big part of the idea here with this AI is to cut that [timeline to clinical trials] significantly,” explains Dr. Joseph Tucker, a drug development veteran and CEO of MagicMed. He says to think of the base structure of the molecule, such as psilocybin, like a key blank. A skillful locksmith—or in this case, biochemist—can develop an almost endless number of permutations.
“We’re trying to get the house key, all the different teeth. Now let’s try them all until you find the right one,” Tucker tells DoubleBlind. The “house” they want to enter is effective treatment for pretty much anything, not just the mental health issues that can be addressed by psilocybin, for example. “We actually start with a target indication in mind. And then we go and say, ‘What receptors are the right receptors? What physiological biomarkers are we looking for in animal models and in humans?’ And let’s try to design a drug out of the Psybrary that best addresses that.”
In early March, MagicMed launched their PsyAI program, recruiting Dr. Eric Nyberg, a professor at Carnegie Mellon University who helped develop the original IBM Watson computer system, which famously won on the gameshow Jeopardy! MagicMed hopes to wield Nyberg’s technology to sift through these millions of chemical designs to find medications that might work like psychedelics for treating a variety of ailments.
How to Grow Shrooms Bundle
Take Both of Our Courses and Save $90!
“We’re essentially moving the molecules much more rapidly by doing stuff in silico,” Tucker says, referring to the Latin phrase for computer simulation. And like other machine learning programs, the more it does its job, the better it gets at finding new drugs. “Comparing it to real data that we generate [through clinical trials], we can improve the predictive capability of this PsyAI, so that it becomes more and more intelligent, more successful in its predictive capabilities.”
Other efforts using AI to generate new psychedelic drugs include the work of molecular pharmacologist Dr. Bryan Roth, at the University of North Carolina-Chapel Hill. As previously reported by DoubleBlind, Roth is was recently awarded $27 million by the U.S. military’s secretive developmental wing DARPA (the Defense Advanced Research Projects Agency) to develop drugs that resemble psychedelics, but don’t cause hallucinations or a trip.
Roth’s lab uses a computational program called Ultra Large Scale Docking (Ultra LSD) that generates millions of permutations of different chemical structures, then predicts how the molecules would fit into the serotonin 2A receptor, which is largely associated with how some psychedelics generate their unique mental effects. LSD, psilocybin and DMT all bind to the serotonin 2A receptor, and Roth has done some of the world’s most groundbreaking research on how these drugs, such as LSD, interact at these receptors.
“We have basically a three-dimensional model of the [serotonin] receptor [where] there are sites for drugs to bind. The computer takes each drug one by one and puts it in there,” Roth explains. If the drug clicks into the serotonin 2A receptor, that’s an indication that it will work like a psychedelic. “The library has been expanded considerably and we’re planning on docking, I think, 5 billion compounds in the summer.”
Roth’s lab has been slowly picking chemicals that seem like promising new medications and working with a lab that then synthesizes them. Finally, a robot tests the drugs by squirting them at human cells to make sure what the computer predicted matches up with reality.
“The project is to make drugs that interact with the same receptor that LSD does, and have the beneficial, putative effects of LSD without having a psychedelic effect,” Roth explains. This is accomplished by picking drugs that have slightly different binding profiles than classic psychedelics, but have removed the downstream effects that cause hallucinations and visuals.
Therefore, these drugs can’t really be called psychedelics. Some in the industry call these new compounds “neurotransformational medicines” because they are a new class of drugs altogether that psychiatry has never encountered before. That raises big questions if they would actually work, but the implications are huge.
Already, Roth and his colleagues have narrowed down a few promising new drugs and begun testing them in mice using a battery of tests that can determine if a drug is psychoactive in animals. For people with debilitating conditions like heart problems or schizophrenia, the powerful sway of psychedelics may not be therapeutic—it may even be harmful. A drug that can ease depression as rapidly as a psychedelic without the introspective distortion would be very useful for certain patients.
But making a drug that looks like LSD without the hallucinogenic aspects is just part of the equation. Roth’s lab still has to determine if these medications will have the same alleged benefits of psychedelics, such as improving mood, erasing suicidal thoughts, or easing the symptoms of post-traumatic stress disorder. However, some argue that the “journey is the medicine” and that without the visual distortions—even “bad” or uncomfortable trips—non-psychedelic psychedelics won’t work as well.
Read: Redesigning Psychedelic Mushrooms to Never Cause a “Bad Trip”
Tucker’s lab will encounter the same issue: It’s nice to generate millions, or even billions, of psychedelic drugs using a computer program, but unless that translates into useful medications, it’s not very practical.
Nonetheless, artificial intelligence and machine learning is rapidly changing drug development, not just for psychedelics, but for the entire pharmaceutical industry. Late last November, DeepMind, (a subsidiary of Alphabet Inc., the company that owns Google), solved protein folding, a longtime puzzle that has plagued biology for around 50 years.
Need Access to Shrooms?
Secure your supply with DoubleBlind!
Protein folding involves using computers to predict a protein’s three-dimensional structure based on its amino acid sequence, which is a lot harder than it sounds. But now that it’s been solved, it will likely accelerate drug development unlike anything previously seen by making it easier to predict how new medications will interact with receptors.
It’s a major step forward in drug development and related efforts in the psychedelic sphere are no less impressive. As Roth previously told me, his project may fail and “we may just end up with better psychedelics.” Better, of course, is subjective.
“Our goal is to ultimately alleviate human suffering,” Roth says. “That’s what we’re doing this for. And I hope we’re successful. It would be wonderful.”
