Please wait while the transcript is being prepared...
0:00
Hello, everyone. My
name is David Clark,
and I'm a Senior Research Leader
at Charles River in the UK.
Today, I'd like to talk to
you about virtual screening
looking back at some
of the experiences
that I've had in this
area during my career,
taking a look at the state
of the art, and also
what might be coming
downstream in the future.
0:24
By way of overview, we'll
begin with a bit of
an introduction to what
virtual screening is,
in case you've never
encountered it.
We'll then have a look
briefly at a couple of
examples that I think illustrate
the state of the art.
I'll then move on to
the bulk of the talk,
which is some of my
personal reflections of
my experiences of virtual
screening during my career.
Then a brief look at some
of the current challenges
that remain in the
field, and how some of
the future directions might
address some of those.
Then I'll finish with
a few conclusions and
recommendations in case you want
to pursue virtual
screening yourself.
1:05
Introduction.
Let's have a look at the
basics of virtual screening.
1:12
In summary, virtual screening
is the rapid assessment,
by computer, of potential
ligands for a protein.
As such, it's the in
silico counterpart
of high-throughput
biochemical screening.
Although we've only been using
the term virtual
screening may be
for a couple of decades or so
this activity has been practiced
in other guises, really
since the beginning of
computer-aided drug design
back in the 1970s.
But it's really in the
last 20-25 years or so
that it's come to prominence,
and there are several
reasons for this.
Firstly, as we're all aware
from our own
computers and phones,
there have been huge advances
in computer hardware
and software in recent
decades, and the graph
on the top of the right
side of the slide shows
that plot of what we
call Moore's law,
showing how the number of
transistors per microprocessor
has continued to increase,
bringing with it
associated increases in
computer performance.
You are also perhaps aware of
the economic pressures
that have been
on drug discovery over
the last few decades.
Trying to bring
drugs to market is
a huge financial and
logistical challenge,
and anything that can be
done to speed that up
and to cut down the
costs is valuable.
Doing things on a
computer is clearly
a lot cheaper than doing
things in a wet lab.
If a virtual screen could be run
instead of or as an adjunct
to a normal screen,
then that could be
hugely beneficial.
A really important factor
is the increase in
the number of 3D protein
structures available,
and the graph at
the bottom right of
this slide shows how the
number of structures in
the protein data bank, which
are publicly available,
has increased over the
last 50 years or so.
You can see that from virtually
nothing around the 1980s,
there's been an almost
exponential rise which continues,
and the collection
actually lodged its
200000th structure
a year or so ago.
So, incredible progress
on that front too.
Something that's happened
even more recently is that
the collections of
compounds available
for virtual screening
have grown enormously.
We've always been used to
having collections
from suppliers of
actual compounds
or compounds with
physical samples that are on
stock and ready to order.
But the real change in
the last decade or so
has been the advent of these
so-called virtual or on
demand databases such
as Enamine REAL,
where the compounds don't
actually physically exist
at the time you look at them.
But if you were to
choose to order them,
they would be able
to be made with
roughly 85% certainty
within 4-6 weeks,
so comparable to ordering
from an on-stock source.
Enamine REAL currently comprises
several billions of compounds.
I've got 6.75 billion
on the slide.
But I think now it's
grown even beyond that.
Those are some of the factors
that have driven the popularity
and the applicability of
virtual screening
in drug discovery.
Virtual screening can
be one of two types.
It can be structure-based,
where we use a 3D model of
the target protein to
screen the compounds,
or it can be ligand-based,
where we use just the structures
of known active compounds
to screen compounds against
them and compare them.
I'm going to cover
each of these in turn.
