166 - Future of Drug Discovery (S12E1)

Welcome to the deep dive. Today we're looking
at something pretty exciting, the future of drug
discovery. It's moving so fast. Right. And you're
probably tuning in because you want to get a
handle on the big shifts, the advancements, shaping
how new medicines get made, but without wading
through super dense academic papers. Yeah, we've
done some of that legwork. We've sifted through
quite a bit, especially looking at studies from
organic process research and development, or
OPR &D as it's often called. Exactly. So our
mission really is to pull out the key forward
-looking ideas the the big paradigm shifts that
are set to change how we find and develop treatments.
And what's really interesting is how different
trends are sort of converging. We've got traditional
pharma research, but now it's being seriously
boosted by, well, digital tech and new ways of
working in the lab. It's not just small adjustments
then. No, not at all. I mean, it feels like a
real sea change in how we approach the whole
complex process of getting new drugs to people
who need them. OK, so let's dive right in. AI,
artificial intelligence, and machine learning.
That comes up constantly. It sounds futuristic,
but it's actually doing real work now, isn't
it? I think I read about something called EVE.
Oh, absolutely. EVE AI is a great example. It
shows how AI can find new uses for drugs we already
have. For e -repurposing. Exactly. Especially
for things like neglected tropical diseases.
Which is huge, because these are areas that often
don't get a lot of funding. Right. Traditional
research might not see the economic incentive
there. Precisely. So AI can potentially find
treatments much more efficiently in these underfunded
areas. The speed is, well, it could be transformative.
That's amazing, finding hidden value like that.
Now, farmer research. It just generates unbelievable
amounts of data, right? Biology, pharmacology,
clinical results. It must be overwhelming. It
really is. Like finding a needle in a haystack,
maybe several haystacks. And managing that, pulling
out useful insights from all that diverse data,
that's a major hurdle. So AI helps with that
data overload. That's where it truly excels.
AI and machine learning can handle these huge,
messy data sets. They can integrate information
even if it's sparse or comes from completely
different experiments. And that allows for something
called cross -dataset analysis. Basically, AI
can spot subtle connections, patterns that a
human researcher might dismiss. This leads to
a much deeper understanding of the disease and
how drugs might work. So it connects dots we
couldn't see before, potentially revealing new
drug targets. Exactly that. And it goes further.
AI can actually help design new drug molecules,
too. From scratch. Pretty much. Algorithms can
look at vast libraries of potential molecules
and predict which ones are likely to hit a specific
target, designing candidates with very specific
properties. Wow. But, and this is important,
AI is a tool. It augments the researchers. It
handles the heavy lifting with the data, finds
patterns. But human creativity, that intuition,
critical thinking that's still absolutely central,
it's a partnership. That makes total sense. Human
ingenuity directing the computational power.
Okay, shifting gears a bit. What about the actual
lab work, the hands -on chemistry? Is tech changing
things there, too? Oh, definitely. Automation
is massive. Think about automated synthesis and
purification laboratories, ASPLs. ASPLs, okay.
Yeah, these are basically robotic setups. They
run chemical synthesis, purification, all guided
remotely by researchers. So a chemist in, say,
London could run an experiment happening in a
lab in San Diego. That's the idea. It could dramatically
boost efficiency, obviously, but also make global
collaboration so much easier. You break down
those geographical barriers. It really does sound
like the future, but also quite practical. OK,
moving along the pipeline, high throughput screening,
HTS. That's testing tons of compounds quickly,
right? That's it. You have these huge libraries
of chemicals, and you screen them rapidly against
a biological target to find initial hits things
that show some activity. Yeah, that must generate.
Just vast amounts of data. Which brings us to
chemoinformatics. You need computational tools
to manage and make sense of all those HTS results.
Chemoinformatics. Right. How does that help specifically
with HTS? Well, after you get those initial hits,
chemoinformatics tools help group them or cluster
them based on their chemical structures, finding
similar molecules. So instead of chasing down
every single hit, which could be thousands, researchers
can focus on maybe one or two representative
compounds from each structural cluster. It's
about working smarter. That sounds much more
manageable. Grouping similar structures lets
you pick representatives. But I guess just looking
at the structure doesn't always tell the whole
story. You've hit on a really critical point
there. Relying only on structure can be misleading
sometimes. You might lump together molecules
that actually behave quite differently biologically
despite looking similar. That's why understanding
the structure -activity relationship SAR is so
vital. How do small changes in structure affect
what the molecule does? Timoinformatics tools
help analyze these SARs so you don't accidentally
throw out a promising candidate just because
it got grouped incorrectly based on a simplified
structural view. Got it. So more nuanced look
than just shape sorting. OK, let's talk manufacturing
quality. Obviously, quality controls paramount.
How are digital tools helping there? Automation
is becoming huge here, too, for both quality
and efficiency. Take Raymond's spectroscopy,
for instance. It's being used more and more for
real time. online measurement of, say, the drug
content while it's being manufactured, like in
extruded drug films. So checking quality during
the process, not just at the end. Exactly. That's
the core idea behind process analytical technology,
or P &T. You build quality into the process by
understanding and controlling it better. Digital
tools are also allowing for much tighter control
over basic steps like granulation, making powder
stick together for tablets. Oh, yes. Better control
there means more consistent, higher quality medicines
in the end. It's proactive quality control baked
right in. Makes sense. What about this continuous
manufacturing idea? That sounds like a big shift
from making drugs in batches. It really is. Instead
of making one batch, stopping, cleaning, then
starting the next, continuous manufacturing uses
integrated systems where everything flows well,
continuously. And the benefit is potentially
much better process control. That usually translates
to consistently high product quality and often
improved safety, too. And regulatory bodies like
the FDA are actually encouraging companies to
adopt these technologies. Interesting. So we
have these big trends, AI, automation, HTS, continuous
manufacturing. But how does this stuff actually
play out day to day? This is where those OPRD
studies you mentioned come in, right? Giving
real examples. Precisely. OPRND is full of case
studies showing these things in action. For example,
you see lots of work on controlling crystallization.
Things like seeding strategies, careful temperature
ramps, choosing the right solvent, all to get
the drug substance in a stable bioavailable crystal
form. Controlling the crystal structure, the
polymorphism. We talked about how crucial that
is before. absolutely critical for efficacy and
safety. And OPR &D papers detail how companies
tackle this for specific drugs like phenofibrate.
outlining exactly how they ensure they get the
right crystal form every time. Real -world problem
solving. Exactly. And you also see loads of examples
of advanced chemistry being used at scale, like
transition metal catalyzed couplings, powerful
reactions for building complex molecules. OPRND
papers often show detailed reaction schemes,
you know, like Scheme 10 .1 or 10 .2 you might
find in some sources, illustrating how these
are used in actual large -scale pharmaceutical
synthesis. So translating cutting edge chemistry
into something manufacturable. What else did
these studies show? Oh, lots on optimization,
refining reaction conditions, making the workup
procedures more efficient, like studies showing
how to drastically reduce the amount of silica
gel needed for purification. That saves money
and is better for the environment. Practical
improvements. Yes, and also novel synthetic methods,
like new ways to do reductive amination, which
is a common reaction type. OPRND publishes these
new improved methods. So it's not just finding
new reactions, but making existing ones better,
cleaner, cheaper. And you mentioned analysis.
the molecules. Definitely. Many OPRND studies
report incredibly detailed characterization of
intermediates the molecules made along the way
to the final drug, using techniques like NMR,
FTIR. To really understand the process. Exactly.
Understanding the reaction pathway deeply. There
are examples where they meticulously characterize
specific intermediates, like a certain carbamate
or puridinium salt, providing that crucial quality
data. Plus, you find completely new synthetic
routes or purification methods for drug candidates,
like preparing a complex molecule called Compound
8A. It sounds incredibly detailed. It is, and
sometimes quite innovative too, like using oxidative
deuromatization to build really complex structures
that were hard to make before. It's amazing to
see that level of detail and innovation in process
chemistry. Now, before manufacturing, there's
preclinical development. Are these tech trends
hitting that stage too? Yes, they are. I mean,
the bedrock is still rigorous preclinical toxicology
studies for safety, of course. We've covered
that before. But digital tools and better molecular
understanding are making an impact. How so? Well,
take CYP inhibition screening. This is done in
vitro in the lab during lead optimization. It
helps predict if a drug candidate might interfere
with metabolic enzymes in the body, causing drug
interactions later on. So you can design out
potential problems early. Exactly. By understanding
how a molecule might inhibit key enzymes like
CYP3A4, which metabolizes many drugs, researchers
can tweak the molecule or select a different
one to minimize that risk way before it gets
near a patient. Proactive safety design. And
what about how the body handles the drug, pharmacokinetics?
Absolutely crucial early on. Things like absorption.
How well does the drug actually get into the
bloodstream? That's fundamental. Medicinal chemistry
teams often face really tricky synthesis challenges
at this stage, sometimes dealing with regioasomers.
Molecules with the same parts just arrange differently.
Right. And getting the right isomer with good
pharmacokinetic properties can be a major hurdle.
Overcoming those synthesis challenges is a big
focus in preclinical work. OK, so it really feels
like the whole landscape is shifting from AI
finding leads to robots in the lab, smart screening,
digital manufacturing. That sums it up pretty
well. All these trends, often highlighted by
the practical research in OPR and D, they're
all aiming for the same thing, making drug discovery
faster, more efficient and ultimately better
medicines. Ultimately, yes. Getting better, safer,
more effective medicines to patients who need
them, that's the goal. This has been really illuminating.
Thinking about all these advancements, AI, automation,
the detailed process chemistry from OPR and D,
it really does make you wonder what's next. What
new doors might open for personalized medicine
or for tackling diseases we currently think of
as well. untreatable. It's a lot to think about,
definitely. The potential is huge, but so are
the challenges and perhaps some ethical questions
down the line. Definitely food for thought. Thanks
for joining me on this deep dive today.