Complexity.

Nothing quite embodies this word
like the human brain.

So for centuries we've studied
the complexity of the human brain

using the tools and technology of the day,

if that's pen and paper
from the age of da Vinci,

through advents in microscopy

to be able to look more deeply
into the brain,

to a lot of the new technologies
that you've heard about today

through imaging,
magnetic resonance imaging,

able to look at the details of the brain.

Now one of the first things
you notice when you look

at a fresh human brain
is the amount of vasculature

that's completely covering this.

The brain is this metabolically
voracious organ.

Approximately a quarter
of the oxygen in your blood,

approximately a fifth
of the glucose in your blood

is being used by this organ.

It's so metabolically active,
there's a waste stream

which comes out
into your cerebral spinal fluid.

You generate 0.5 liter of CSF every day.

So, as you know, researchers
have taken advantage

of this massive amount of blood flow
and metabolic activity

to begin to map regions of the brain,
to functionally annotate the brain

in very meaningful ways.

You'll hear a lot more
about those kinds of studies,

but basically taking advantage of the fact
that there's active metabolism

with certain tasks going on.

You can put a living human in a machine

and you can see various areas
that are lighting up.

For example, going around right now
is the temporal cortex,

auditory processing going on there,
you're listening to my words,

you're processing what I'm saying.

Moving to the front of this brain
is your prefrontal cortex,

your executive decision-making,

your higher-thinking areas of the brain.

And so the thing that
we're very much interested in

from the perspective
of the Allen Institute

is to go deeper,
to get down to the cellular level.

So when you look at this slice, it doesn't
really look like gray matter, does it?

It's more tan matter, or beige matter.

And scientists about, I guess
around the late 1800's,

discovered that they could
stain tissue in various ways,

and this sort of came along
with various microscopy techniques.

And so this is a stain, it's called Nissl,
and it stains cell bodies,

it stains the cell bodies purple.

And so you can see
a lot more structure and texture

when you look at something like this.

You can see the outer layers of the brain
and the neocortex,

there's a six-layer structure, arguably
what makes us most uniquely human.

As you've heard before about
there's on average in a human,

there's about 86 billion neurons,
and those 86 billion neurons

you can see are not evenly distributed,

they're very focused
in specific structures.

And each of them
has their own sort of function,

both on an anatomic level
and at a cellular level.

So if we zoom in on these cells,
what you can see is large cells

and small support cells
that are glials and astrocytes

and these cells are as we know
connected in a variety of different ways.

And we like to think about,
although there's 86 billion cells,

each cell might be considered a snowflake,
they're actually able to be binned

into a large number
of cell types or classes.

What flavor of activity
that particular cell class has

is driven by the underlying genes
that are turned on in that cell,

those drive protein expression
which guide the function of those cells,

who they're connected to,
what their morphology is,

and we're very much interested
in understanding these cell classes.

So how do we do that?

Well, we look inside the cell
at the nucleus,

-- and it will get to the nucleus --

and so inside we've got
23 pairs of chromosomes,

got a pair from mom, a pair from dad,

on those chromosomes about 25,000 genes
and we're very much again interested in

understanding which
of these 25,000 genes are turned on

at what levels they're turned on.

Those are going to, of course, drive
the underlying biochemistry of the cells

they're turned on in and again every cell
in our bodies more or less has these

and we want to understand better

what the driving biochemistry
driven by our genome is.

So how do we do that?

We're going to deconstruct a brain
in several easy steps.

So we start
at a medical examiner's office.

This is a place
where the dead are brought in

and obviously as you saw before

for the kind of work we do,
[it] is not non-invasive,

we actually need
to obtain fresh brain tissue

and we need to obtain it within 24 hours
because the tissues start to degrade.

We also wanted for our projects
to have normal tissue,

as much normal as we could possibly get.

So over the course of a two-
or three-year collection time window

we collected 6 very high-quality brains,
5 of them were male, 1 was female.

That's only because males
tend to die untimely deaths

more frequently than females,
and then to add to that,

females are much more likely
to give consent

for us to take the brain than vice versa.

We have to figure that one out.

We've heard people say,
"He wasn't using it anyway!"

(Laughter)

So, once the brain comes in
we have to move very, very quickly.

So first we capture
a magnetic resonance image.

This, of course,
will look very familiar to you,

but this is going to be the structure
in which we hang all this information,

it's also a common coordinate framework
by which the many, many researchers

who do imaging studies can map

into our ultimate database,
an Atlas framework.

We also collect diffusion tensor images,

so we get some of the wiring
from these brains.

And then the brain
is removed from the skull.

It's slabbed and frozen solid,
and then it's shipped to Seattle

where we have
the Allen Institute for Brain Science.

We have great technicians
who've worked out

a lot of great techniques
for further processing.

So first, we take a very thin section,
this is a 25 micron thin section,

which is about a baby's hair width.

That's transferred to a microscope slide
and then that is stained

with one of those histological stains
that I talked about before.

And this is going to give us more contrast
as our team of anatomists

start to make assignments of anatomy.

So we digitize these images,

everything goes from being
wet lab to being dry lab.

And then combined with anatomy
that we get from the MR,

we further fragment the brain.

This is to get it into a smaller framework
for which we can do this.

So here's a technician
who's doing additional cutting.

This is again a 25 micron thin section.

You'll see da Vinci's tools,
the paintbrush,

being used here to smooth this out.

This is fresh frozen brain tissue.

And it can be very carefully
melted to a microscope slide.

You'll note that there's a barcode
on the slide.

We process 1000's and 1000's of samples,

we track all of it in a backend
information management system.

Those are stained.

And then we get
more detailed anatomic information.

That information...

This is a laser capture microscope.

The lab technician is actually describing
an area on that slide.

And a laser, you see the blue light
cutting around there,

very James Bond-like,
cutting out part of that.

And underneath there,
you can see the blue light again,

from the microscope in real-time,

it's collecting,
in a microscope tube, that tissue.

We extract RNA,

RNA is the product of the genes
that are being turned on,

and we label it,
we put a fluorescent tag on it.

Now what you are looking at here

is a constellation
of the entire human genome

spread out over a glass slide.

Those little bits are representing
the 25,000 genes.

There's about 60,000 of these spots
and that fluorescently labeled RNA

is put onto this microscope slide
and then we read out quantitatively

what genes are turned on at what levels.

So we do this over and over and over again
for brains that we've collected;

as I mentioned we've collected
6 brains in total.

We collect samples
from about 1000 structures in every brain

that we've looked at,
so it's a massive amount of data.

And we pull all of this together,
back into a common framework,

that is a free and open resource
for scientists around the world to use.

So at the Allen Institute
for Brain Science,

we've been generating these kinds
of data resources for almost a decade.

They're free to use for anybody,
they're online tools,

just for example today a given workday,
there'll be about 1000 unique visitors

that come in from labs around the world,
to come use our resources and data.

They get access to tools like this,
which allows them to see

all of that anatomy and the structure
that we created before

and to start mapping in then the things
that they're particularly interested in.

So in this case you're looking
at the structure

and they're going to look
at these color balls

are representing a particular gene
they're interested in

that's either being turned up or down

in those various areas depending upon
the heat color that's specified there.

So what are people doing
when they start using these resources?

Well, one of the things
that you might hear lots about

is human genetic studies.

Obviously, if you're very interested
in understanding disease

there's a genetic underpinning
to many of them.

So you'd like more information,
you do a large-scale study

and you get out of those studies
collections of genes

and one of the first things you're going
to want to know is more information.

Is there something I can learn
about the location of these genes

that gives me additional clues
as to their function,

ways in which I might intervene
in the disease process.

They're also very interested
in understanding human genetic diversity.

We've only looked at 6 brains,

but, as we know,
every human is very unique.

We celebrate our differences;

this is a snapshot of the great workforce
at the Allen Institute for Brain Science

who does all the great work
that I'm talking about today.

But remarkably when we look at this level
at the underlying data,

and this is a lot of data from
2 completely unrelated individuals,

there's a very high degree
of correlation, correspondence.

So this is looking at thousands
of different measurements

of gene expression across
many, many different areas of the brain;

and there's a very high degree
of correspondence.

This was very reassuring to us.

First, because when you generate
data on this scale,

you want to make sure it's high quality,

so reproducibility is obviously important,

but it was also important
because we feel that it's given us

a great snapshot into the human brain.

And the people using the data,
even with our low N, have confidence

that what they're seeing
has some relevance.

Now, not everything is correlated here,
you can see some outliers,

and, of course, those outliers
are going to be interesting

related to human differences.

We did a study a couple of years ago,

in which we tried to understand
a little better about those differences,

and looked at multiple individuals
and different gene products,

and what we find, as a tendency
and as a rule,

is that those differences tend to be
in very specific cell populations

or cell types, cell classes,
as I mentioned before.

So, this is an example
of 2 different genes that are turned on

in very specific layers of the neocortex

only in one individual
and not found in another.

Now we have no idea
if that's due to environmental changes,

environmental influences
or if it's just genetics,

but we did do a study in which we looked
at the mouse several years ago

and we were looking at genes
that encode for, in this case a DRD2,

the gene listed on the top
is a dopamine receptor.

Tyrosine hydroxylase, TH, is a gene
involved in dopamine biosynthesis

and those 2 gene products
are very different in the cell types

in these individual mouse brains.

So, over on the left is "C57 Black 6"
which is a commonly used mouse strain,

and then spread at the other end
is a wild type strain.

And so the further you go
the more genetically unrelated you are.

And when we looked in total across,
sort of evolution if you will,

across genetic relatedness,

the further you were
genetically unrelated,

the more of these
very specific cell types,

specific changes, you could see.

So at the Allen Institute
for the next decade

we're embarking
on a pretty ambitious program

to start to understand the cell types,
understand the cell differences

and how they ultimately relate
to the functional properties of the brain.

This is, I think, critical information
for the entire field,

to start linking up all
of these fundamental parts

which are the cells,
to how they're connected,

the underlying molecules
that drive those connections,

the underlying molecules driving
the electrophysiological properties,

the electrochemical properties

and then ultimately
the functional properties of those cells.

So we're doing this
in 3 different areas of research.

First, we're focusing on the mouse,
the mouse visual system,

to look at, in real-time,
in the living animal,

the functions of a variety
of different cells.

We're linking these in this concept
in the middle of cell types,

trying to really understand
the underlying molecules

in all the properties
as they relate to those functions

and then we're looking at the human.

In the human we're doing this both
in the middle and cell types

using the tissue driven work
that I talked about before,

but also we're doing it in vitro
using stem cell technology.

We're learning how to make
very specific cell types within the dish

and then being able to test
those functional properties

and go back and forth between
what we learn in the mouse to the human.

So, with that I will finish
and just say that it's an exciting time

to be in biology and an exciting time
to be in neuroscience.

I think the technology of the day
has come well beyond the pen and paper

and it's really time for a renaissance in
our understanding of this complex organ.

Thanks.

(Applause)