One Scientist. A Billion Experiments.

One Scientist. A Billion Experiments.


I have a story to tell. It’s a story about drug discovery and about finding the right chemical compounds
to fight disease. How many compounds are out there with the
potential to fight disease? Literally … trillions. We begin with … aspirin. Aspirin is a drug we’ve known about for hundreds of years. It’s a chemical compound — a molecule — made
of carbon, oxygen, and hydrogen atoms. It works, in part, by binding to and disabling
target proteins in our bodies. That’s basically how small-molecule drugs
work – by binding to proteins that play a role in
disease. So how do we replicate aspirin’s success and find compounds that will bind to disease-enabling
proteins? How do we find the therapeutic needle in the
haystack? Let’s say we have a protein that helps cancer
cells grow. There’s no way we could test, by experiment, every compound that might block the activity
of that protein. There just aren’t enough labs and isn’t
enough time. But we could use computers to rapidly screen many millions of compounds, in silico. We’ve built a system that does exactly that. It’s called VirtualFlow. With VirtualFlow, we begin by preparing both
the protein, and the chemical compounds. For a compound, this means taking very basic
information about its atoms and how they’re connected and turning it into a 3D model. VirtualFlow uses as many as 100,000 computer
processing cores to prepare many millions of compounds. So far, we have created models for 1.4 billion
compounds that are commercially available. But it could do more. VirtualFlow also runs “docking” programs. Docking programs bring together the protein
of interest with each compound. The program then checks whether the compound
is able to bind to the protein. Each compound is able to take on many shapes. The program tests many thousands and it gives each shape a “stickiness”
score, depending on the likelihood that it will attach
to the protein. This is repeated hundreds of millions of times,
in parallel, by all of those computer cores. VirtualFlow then takes the stickiest compounds
— the top contenders — and retests them, but this time with greater
precision. With our list of likely candidates — typically
about a hundred — we leave the virtual world, go to a lab, and test the actual compounds with actual
proteins. Traditional real-world screening methods test only hundreds of thousands of compounds per
experiment. VirtualFlow can test a billion or more, and it can do it faster and for far less cost. That’s my story. Well, I guess you could say that it’s just
the beginning of my story. So which compound will turn out to be the
next aspirin? I can’t wait to find out.

Comments

(7 Comments)

  • Codelexx

    This is oddly unsettling

  • Random Completely

    Nice video

  • TheComicalCanadian

    great video keep up the good content

  • TheComicalCanadian

    nice video bro

  • TheComicalCanadian

    amazing content keep it up dude

  • A., Miles C.

    What's with the Echo And The Bunnymen outfit and Page Boy haircut? Is this Science or pretentious 80's Performance Art from NYC?! 😆😆🤣🤣🤣🤣

  • Lyle Gorch

    This is a promo for the new Norse Noir series on Netflix.

  • Your email address will not be published. Required fields are marked *