AI & Tech Use in Drug Discovery

What are some current challenges during the earlier phases in the drug discovery process?

Current challenges are not just ones we're facing today, but they've kind of been accumulating over the past decades. First and foremost, biology. Biology is really hard when we're talking about understanding the immune system. And we come to grips with the fact that, you know, just the immune cell, that's that what we call B cells. I mean, the diversity of those cells is vast, I mean, it's anywhere from 10, to the 11th, to 10, to the 15, different type of immune cells, B cells per person, right. So that's a big one, different genetic backgrounds for people, not just individuals, but across populations.

Different environmental exposures, you know, some folks, you know, may work in a coal mine, or in a factory where formaldehyde is off gassing, you know, we're all exposed to things every day. The data, I guess that's probably a really big one, too. We have a lot of different data types that we need to make sense of that we need to integrate, harmonize relate together, that's really difficult. business practices are always a big issue. So, it's, it's a multifaceted complexity that I think makes drug discovery quite difficult.

How is artificial intelligence currently being used to assist in drug therapy research?

A number of different ways. Primarily, one of the one of the ways is drug repurposing. So, drugs that have already been approved, for certain indications, AI is being used to look at those drugs, in different lights for different applications. So that's a really big one. And that, that gets through the FDA clinical trial process, because they've already come through. And they've been shown to be successful, right. Target Discovery, so looking for new targets, within us within our human system, that's a really big one, our bodies are really complex. AI is fantastic for tackling complexity, right, I would say, I would say optimizing, you know, what we've already done in terms of clinical trials. So, you know, when you design a clinical trial, there are a lot of different facets to do that with.

AI, I think, is fantastic to help us, you know, less than to understand which knobs to turn, when putting a clinical trial together to look at a given condition. That's a really big one. Probably drug optimization is another big one. So okay, we already have a drug, we have a target, we know how it binds to the target. But are there different ways we can tweak that drug so that it binds better or longer, right? These drugs are in three-dimensional space. And they're not just the static molecules. And so, talking about antibody therapeutics, you know, in a three-dimensional space, they kind of wiggle around a little bit and their targets wiggle around a little bit. And both of these two things wiggling around and then binding together and wiggling around together. So, there's quite a bit of complexity there. Those are probably some of the biggest that I can think of.

What technology, in addition to AI, is being used to shorten the drug discovery funnel and reduce research costs?

So, in R&D, there are lots of different types of hardware that look at different facets of, of how our immune systems respond, okay? And I'll just name three of the larger ones right off the bat. So, there's sequence-based analysis. So genetic sequence DNA RNA protein sequences, that's one space and other spaces proteomics. So really the study of those proteins what are they comprise of? How are they? How are they characterized another third large business flow cytometry, so where machines are used to characterize immune response to look at different cell populations, right? If a patient is, is in a drug trial, and we want to see how their immune system is responding to the drug, then we can use a flow cytometer to look at how the immune system is responding.

So, those are three main data streams and really, I think where AI is going to help is it's going to allow us to kind of coalesce these data streams or layer them or integrate them together, make lots of many too many comparisons. Then we can come up with different hypotheses, more so more complex than what we could have. So that's, I guess, kind of more broadly speaking, but specifically, I mean, we have things like high throughput sequencing, right, so we can generate, you know, gigabytes worth of sequence data per person, right, we have CRISPR, cast nine systems that are, you know, have been in the media for quite some time. But those are gaining traction, as well as cell and gene therapies.

Those are really coming into maturity now, especially when you're looking at blood diseases like sickle cell anemia, or beta thalassemia. Those are starting to really gain maturity. Automation, and robotics is a really big one, right? So that, you know, you have hardware that can work 24 hours a day, right. Whereas, you know, we humans get tired, we have limitations, we have families, we need to go home and get rest, right? Automation is a really big deal. I was just at an automation conference a couple of weeks ago called SLA s. And there you see the cutting edge of automation technologies from, from sample handling to sample tracking to you know, putting together different experimental designs. And I think probably the biggest one for us at dogmatics. Is platform, right? The software platform that can act as the hub that everything hooks into. So there's a lot of really interesting work that's going on in that space.

I think other pieces of hardware, I would say GPUs, which is, you know, a piece of hardware graphical processing unit that uses that technology to speed up calculations, that's a big one. I'm trying to think silicon chips, there's always a race to have better, faster silicon chips, right? So really, we've got kind of a blend of hardware and software, and then biotechnology is kind of all coming together.

What does this mean for faster and better treatments overall?

I want to be careful here, they're faster doesn't always mean better, right? So, you can, you can make mistakes faster. And that doesn't mean you're doing a great job. However, the way we see it is, if we can throw out the dead ends quicker, then we can focus on those things that have Promise. Okay. So, if you think of the drug discovery funnel, which can be seen, as you know, it's wider at the top, we have lots of things to explore at the beginning. And then as the R&D process moves forward, the funnel gets thinner and thinner and thinner. Right? So, we're taking that same paradigm, and we're actually broadening the top of that funnel quite far. Right? So, we're expanding the number of targets, and we're reducing that target space pretty fast. Okay, so we're able to generate, I think, better hypotheses faster. Right. So, one of the analogies I use is you can build a house with a hammer, not a problem, right? It's going to take you a while. But if you had a nail gun, you're doing the same job. And you're doing it faster. Right. And so that's the way we look at AI. So, it's kind of a force multiplier, right, you're able to multiply your efforts, have more shots on goal, hopefully. And that really helps us you know, navigate through this complex space. Better right.

Now in terms of better treatments, you know, I'll just kind of point to that drug repurposing point I made earlier, if you're able to make small tweaks on a three-dimensional protein, which cause it to last a little longer or bind a little tighter, that's, that's a win for everybody. Right. And so, when we say better, you know, there are a lot of different ways to qualify that. So, it really means us doing more work more efficiently and more intelligently guided.