►
From YouTube: DevoWorm #36: Diatom Dynamics, Biological Tensegrity, Nervous System Structurogenesis in Organoids
Description
Diatom dynamics and image processing, biological tensegrity in tissues and ways to assess stress and strain in embryos. Self-assembly and physical forces in cells/tissues. Formation of nervous system structures (structurogenesis) in mouse and human-derived organoids. Attendees: Bradly Alicea, Richard Gordon, Susan Crawford-Young, and Alon Samuel
B
A
A
A
A
Just
updating
on
the
I'm
going
to
be
doing
lately.
Sorry
regarding
the
dad
Thomas
in
the
Boston
area.
Okay
and
yes,
I
took
a
video
trimmed
like
a
segment
of
it
to
kind
of
track,
because
it's
moving
a
bit
fast
and
like
it's
a
bit
like
zoomed
in
so
it
was
just
a
second
or
a
second
and
a
half
did
create
a
new
video
I
Decided,
like
a
tracking
point
with
on
the
diatom
I'm
going
to
show
it
in
the
in
a
second
I
can
have
an
image.
A
Okay,
I
activated
the
tracker
to
track
this
point,
I
kind
of
showed
on
the
image
like
the
tracking
path.
I
got
the
pixels
and
I
saved
it
to
file,
and
there
are
still
more
stuff
that
I
wrote
for
myself
to
do
before.
I'm
gonna
send
it
to
Thomas.
So
he
sent
me
a
scaling
image
so
kind
of
like
on
the
image
of
the
movie.
A
A
B
A
Yeah,
oh
okay,
yeah,
so
you
can
see
like
the
line
like
the
tracking
line,
like
it
start
like
from
here.
That's
the
big
beginning
of
the
movie
and
then
like
it
goes,
that's
the
end
of
the
movie
and
it's
kind
of
like
because
it's
just
a
second
and
a
half,
because
the
diatom
goes
from
here
and
then
to
kind
of
finishes
like
already
like
here.
It's
just
in
the
end
of
the
field
of
view,
and
we
have
to
get
a
tracking
video
again,
it's
very
short
but
we're
making.
A
B
B
B
A
D
Quite
like
east
east,
with
the
calculation
that
Thomas
did,
don't
you
think
that
it's
going
to
help
like
deciding
if
the
camera
itself
is
jerky
or
like
just
I,
don't
I,
don't
know
the
problem.
The
problem
is
to
calibrate
a
camera
for
jerkiness.
If
you
need
a
smooth
motion
and
we're
trying
to
look
at
a
smooth
motion,
so
you
have
to
have
something
that
a
priority,
you
sure,
is
a
smooth
motion
and
get
this
scale.
The
only
thing
that
you
can
be
sure
is
the
smooth
motion
would
be
Stokes
law
of
an
inanimate
object.
D
D
D
D
I
I
mean
Einstein
used
Stokes
law
for
very
small
particles
to
relate
the
to
demonstrate
that
molecules
existed
with
Brownian
motion.
It
was
probably
emotion
right,
okay,
but
for
large
particles
you
don't
have
to
worry
about
it,
so
large
large
can
be
typically.
Yes,
let's
say
a
millimeter
diameter.
You
could
drop
a
millimeter
bead
of
density,
greater
than
water
in
water
and
it
will
accelerate
and
then
go
at
a
constant
speed.
D
C
D
C
D
D
Okay,
so
you
know
I,
guess
Appleton
designed
the
camera
for
me,
but
okay
and
anyway,
so
so
a
camera
you
have
to
calibrate
the
camera
itself.
That's
the
only
way
I
can
think
of
doing.
That
is
something
who's
smoothly.
Stokes
law
gives
you
the
opportunity
to
do
that,
but
Thomas
may
be
upset
trying
to
turn
his
his
microscope.
90
degrees.
D
D
C
D
D
D
A
D
B
A
A
A
D
A
B
A
D
Edge
detection
and
yeah
Edge
detection
and
then
get
getting
the
angles
of
the
edges.
Yeah
angle
prediction:
yeah:
yeah,
because
that
that
will
tell
you
the
angle,
then
you
can
rotate
it
and
you
should
be
done
on
now.
We're
making
an
assumption
here.
That
is
that
all
the
diatoms
in
the
field
of
view
have
are
parallel
to
each
other.
So
if
you
use
Edge
detection,
you
could
also
check
that
assumption.
D
B
We
did
something
similar
to
this
a
couple
years
ago,
when
we
were
working
with
some
other
images,
we
used
sort
of
a
line
approach
to
align
all
the
diatom
colonies
in
a
single
Direction
so
like
for
the
we're
doing
deep
learning,
pre-training
of
a
model,
so
you
know
you'd
need
them
all
in
the
same
orientation
to
get
a
good
training
set,
and
so
it's
just
the
matter,
but
this
was
just
simple
rotating
around
a
line.
There,
isn't
you
know
I,
think
polar
coordinates
would
be,
maybe
a
bit
better.
B
Just
because
you
know
you
have
well
I,
guess
you're
talking
about
around
a
circle,
yes
yeah!
So
that's
that's
pretty
much
yeah.
That
would
help
I
think
to
make
it
sort
of
a
line
because
you're
actually
not
dealing
with
a
straight
line.
I
mean
the
colonies
as
you
see
they
they
fold,
but
they
have
like
this.
You
know
they
fold
in
a
certain
direction
and
they
they
form.
A
B
A
B
A
Yeah
yeah,
maybe
it's
if,
if
you're
saying
that
that's
probably
this
kind
of
like
just
this
module,
gonna
be
a
good
module
kind
of
further
on.
So
maybe
we're
just
gonna
make
it
like
within
a
different
file
than
just
write
a
bit
of
documentation
for
it
that
people
can
use
them
in
the
future.
Yeah
foreign.
B
B
D
B
So
you
know,
Diva
learn
is
our
where
we
do
a
lot
of
our
machine
learning
activities,
but
we
also
have
data
science,
demos
and
we
have
now
Devo
graph
as
part
of
this.
So
this
has
been
cloned
from
where
we
were
doing
the
Google
summer
code
work
and
we've
put
it
onto
this
space.
So
now
we
have
a
space
for
devograph,
it's
not
in
the
formal
release
of
Divo
learn.
B
B
So
we
can
make
another
release
or
I
guess
we
could
make
a
separate
release
like
a
module
for
devograph
and
we'll
see
what
happens.
I'll,
try
to
figure
out
what
the
best
course
of
action
is.
But
this
is
now
cloned
here
in
Diva.
Learn
we
haven't
done.
I
haven't
done
anything
with
respect
to
the
digital
microsphere
stuff,
yet
that's
kind
of
still
I,
don't
know
whether
I'm
going
to
add
it
into
diva,
learn
or
not.
B
B
B
B
Are
some
issues
earlier
with
it
working
or
not
working,
then
they
moved
to
a
payment
plan-
I
don't
remember,
but
it's
it
should
still
be
okay
for
casual
use.
If
you
really
want
to
do
something
intensive
with
it,
I
download
the
version
that
you
install
on
your
machine,
but
we
have
the
version
up
on
a
Roku
app
too
for
people
mainly
to
like
get
used
to
it
so
cool.
B
D
B
B
Let
me
zoom
in
and
let
it
text
so
the
abstract
reads:
cell
generated
forces
play
a
foundational
role
in
tissue,
Dynamics
and
homeostasis
and
are
critically
important
in
several
biological
processes,
including
some
migration
wound
healing
orphogenesis
and
cancer.
Cancer
metastasis
quantifying
such
forces
and
Vivo
is
technically
challenging
requires
novel
strategies
that
capture
mechanical
information
across
molecular
cellular
and
tissue
length
scales,
while
allowing
these
studies
to
be
performed
in
physiologically
realistic
biological
models,
so
Advanced
biomaterials
can
be
designed
to
be
non-destructively
designed
and
non-destructively
measure.
B
D
B
D
B
A
B
Is
multiple
length
scales,
but
also
these
different
modes
of
function?
I
guess
so
they
kind
of
go
through
they
talk
about
the
human
body
and
how
the
human
body
is
a
dynamic
and
self-stabilizing
structure.
Brings
your
intricate
connections
between
hierarchical
building
blocks,
so
I
guess
the
hierarchical
building
blocks.
Are
these
parts
of
the
you
know
things
that
make
up
the
body?
So
you
know
you
have
cells
that
become
tissues,
tissues
that
become
different
parts
of
the
body
and
so
forth,
and
they
build
from
there
and
the
they
say.
B
So
the
mechanical
structure
of
intron,
extracellular
proteins,
for
example,
cells
and
tissues,
play
a
key
role
in
achieving
structural
stability
in
response
to
widely
varying
mechanical
challenges.
So
again,
this
is
where
you
have
a
lot
of
different
things
that
the
body
needs
to
do
and
those
challenges
are
met
by
the
sort
of
stability.
So
it
does
yeah.
It
does
seem
like
it's
kind
of
like
you
know.
In
a
sense,
it's
useless
because
it.
B
B
B
D
D
C
D
C
D
D
D
B
A
A
D
B
B
C
D
C
B
C
D
C
C
D
D
B
B
D
C
C
B
B
C
D
B
If
we
go
back
to
the
paper,
they
talk
about
biomaterials
and
how
these
can
be
used
in
different
ways
and
that
if
you
have
cell
forces
so
generated
forces
playing
a
critical
role
in
a
lot
of
functional
things
like
tissue,
morphogenesis,
muscle,
contraction,
wound
healing
and
then
dysregulation
of
these
forces
often
correlate
with
disease
onset
and
progression.
So
they're
looking
at
this
kind
of
usefully
in
terms
of
the
sort
of
dysregulation
of
forces,
and
so
let's
see
if
they
have
any
images
here.
B
So
this
is
a
cell
where
you
have
tension
acting
on
a
bunch
of
cells,
pushing
them
together
as
sort
of
like
a
focal
point,
and
then
you
have
that's
actually
compression,
and
then
you
have
tension
operating
on
sort
of
the
rotational
orientation
of
the
cells
operating
side
to
side.
So
if
you
see
these
cells
as
they
migrate
around
in
this
mass
of
cells,
they
experience
may
be
both
tension
and
compression
depending
on
their
location
in
the
in
the
cell
body.
Here
and
then,
if
you
look
at
the
cell,
what
was
that.
B
B
Then,
if
you
go
to
a
single
cell,
you
see
that
there
are
these
filaments
inside
the
cell.
You
have
microtubules
and
you
have
these
same
forces
acting
on
the
cell.
So
it's
not
just
this
population
of
cells
and
again,
if
you
can't,
you
can
apply
for
or
cell
generated
forces,
can
Inspire
to
force
certain
differentiation
programs.
So
there's
a
lot
of
experiments
with
using
cell
forces
or
you
know,
putting
cells
on
different
meat
on
different
substrates
and
they
differentiate
in
different
ways.
B
D
Now
that
you
can
do
with
much
of
our
embryo
data,
you
take
two
cells
and
consider
the
distance
between
them,
and
if
that
changes,
you
could
call
that
a
change
in
string
yeah.
Well,
the
interesting
question
strain
is
easy
to
measure.
In
other
words,
if
you
can
get
the
coordinates
of
the
centroids
or
the
components
right,
for
instance,
cells
in
the
embryo,
the
interesting
question
is
whether
or
not
there
is
a
relationship
between
stress
and
strength.
Stress
is
the
force
between
them.
C
D
C
D
Oh
okay,
neat,
okay,
so
I'm
suggesting
then
that
they,
it
may
be
time
to
move
on
and
get
the
stress-strained
relationships
in
embryos
and
see.
If
we
can
make
anything
sensible
great,
you
could
certainly
get
the
strength
and
the
string
strain
maps.
You
can
take
a
nematode
and
take
any
pair
of
cells,
draw
a
straight
line
between
their
centroids
and
see
how
the
length
of
that
chain
of
that
straight
line.
Changes
over
time
points.
B
B
C
C
D
B
B
D
Well,
there's
also
you
go
back
to
the
old
experiments
by
Steinberg
and
moscona,
where
they
simply
took
two
embryonic
tissues
made.
What
we
would
Now
call
an
organized
can't
watch
the
the
change
of
shape,
such
as
into
an
onion
shape,
where
one
tissue
coated
the
other.
Oh
yeah,
okay,
okay,
I
I've,
been
following
some
webinars
and
a
group
on
soft
soft
science
yeah.
C
B
So
this
is
yeah.
This
is
some
more
examples
here
of.
Actually,
this
is
where
they're
using
molecular
techniques
for
cell
measuring
cell
generated
forces
and,
in
this
case,
they're
using
things
like
fret,
which
is
a
molecular
technique
to
look
at
forces
and
they're
wanting
to
look
at
like
specific
things,
pro
cell
processes,
for
example,
and
they're
measuring
them
this
way.
B
So
there
are
different
ways
you
can
measure
this
forces,
you
can
look
at
this
whole
cells
or
you
can
look
at
parts
of
cells.
So
it's
very
you
know,
then
you
have
these
kind
of
different
types
of
probes,
so
they
have
protein
probes,
which
are
the
Fret
protein
fusions
and
then
DNA
probes,
which
are
DNA
oligonucleotides.
They
you
can
use
in
a
similar
way
and
it
just
measures
local
forces
in
some
parts
of
the
cell.
B
Yeah
I've
seen
I've
seen
a
lot
of
people
do
stuff
with
fret,
like
with
whole
cells.
You
know
like,
if
you're
just
trying
to
get
a
handle
on
like
how
cells
are
behaving
and
you
don't
have
like
you
know,
high-end
equipment,
or
you
want
to
look
at
it
in
bulk.
That's
actually
a
pretty
good
technique
for
it,
but
anyways.
So
this
is
where
you
also
looked
at
things
at
the
cellular
length
scale.
These
are
a
couple
of
different
indicators
here,
so
you
can
adhere
cells,
a
silicone
rubber,
film
and
then
wrinkle
the
substrate.
B
So
like,
as
I
said,
you
can
put
cells
on
a
substrate
and
you
can
play
with
the
substrate.
You
can
play
with
the
shape
or
the
orientation
of
it,
and
you
can
get
the
cell
to
do
things.
You
can.
Actually,
you
can
measure
the
forces,
but
you
can
also
introduce
forces
to
the
cell,
so
this
is
where
the
substrate
is
wrinkled.
In
this
case,
this
is
where
the
substrate
has
like
sort
of
a
stippling
or
something
where
the
cell
is
moving
across.
It
yeah.
D
Bradley
he
seem
to
have
missed
patterned
substrates
yeah.
There's
all
literature
on
I
think
English
may
have
started
it
where
you
make
a
surface
which
has,
for
instance,
says
checkerboard
of
hydrophobic
and
hydrophilic
regions.
Okay
and
I.
Remember:
Ingrid
got
the
interesting
result
of
the
cell.
If
you,
if
you
made
a
square,
if
you
made
squares,
the
cells
would
become
Square.
Oh.
B
A
B
Micropost
then
they
have
they
kind
of
get
into
some
of
these
3D
matrices,
which
is
you
know,
moving
closer
to
the
tensegrity
structures
that
we
were
looking
at
with
the
toy
models
where
you
have
things
in
three
dimensions
because
measuring
things
in
two
Dimensions,
it's
fine
for
measuring
forces.
If
you
want
to
be
really
reductionist
about
it,
but
in
3D,
of
course,
is
where
a
lot
of
these
structures
live.
So
you
can
do
a
lot
of
things
with
traction
forces
in
three
dimensions.
B
You
can
use
collagen,
februin
or
macro
matrogel
matrices
to
sort
of
do
the
same
things
that
you
do
in
2D,
I'm.
Looking
for
more
images,
here's
something!
So
this
is
a
figure
techniques
for
measuring
tissue
scale
forces.
So
now
that
we're
moving
up
to
the
tissue
scale,
where
we
have
a
bunch
of
cells
in
a
population
and
again
we
have
our
our
different
types
of
measurement
techniques.
So
the
arrows
indicate
deformations.
You
have
the
cellulating
matrices
in
a
where
you
have
cell.
B
B
So
this
is
almost
like,
like
the
maybe
the
Leaf
of
a
cell,
that's
coming
out
from
a
bud,
so
this
is
like
unrolling
or
rolling
up
along
this
axis,
so
they're
Contracting
or
expanding
along
this
axis,
and
it
looks
like
they're
curling,
like
this
they're
all
on
this
purple
substrate.
So
that's
why
you
get
that
kind
of
shape
and
then
f
is
embedded.
Micro
cantilevers
include
pre-stress,
with
the
self-assembled
tissue
constructs
and
the
deflection
of
cantilevers
are
correlated
with
contractility.
B
You
get
this
pattern
formation,
and
so
this
is
a
these
are
examples
down
here:
H
through
n,
where
you
have
different
forces
acting
on
the
cell
population.
You're
starting
to
get
different
patterns
emerge
from
those
interactions,
and
so
then,
let's
see
I,
don't
know
if
there
are
any
other
figures
in
here.
I
think
there's
one
more
this
one
here.
This
is
where
you
have
the
10
security
Spectrum.
So
these
are
your
stable,
Dynamic
and
unstable
structures.
So
you
have
this
this
scale
at
the
right.
B
It
says
tension
versus
compression,
so
the
purple
is
where
things
are
in
tension
and
the
red
is
or
they're
in
compression,
and
you
have
these
structures.
So
you
start
with
a
sphere
okay
and
then
you're
going
to
deform
that
sphere
in
different
ways
to
you
know
by
introducing
forces
and
you're
going
to
change
the
shape
of
it,
but
you're
also
introducing
tension
and
compression
at
different
points
in
that
structure.
B
So
you
can
see
that
in
this
sphere
it's
all
compression
or
it's
you
know,
kind
of
neutral,
and
then
you
start
taking
attention
along
the
edges
but
compression
in
the
middle,
and
then
that
makes
like
kind
of
a
goes
from
a
sphere
to
sort
of
an
oblong
shape.
So
it
almost
looks
like
an
egg
where
you
have
an
oblong
shape,
it's
kind
of
like
C
elegans.
We
were
talking
about
where
you
can
deform
the
this
spherical
shape
in
different
ways
and
see
elegans.
B
D
C
B
B
C
B
B
A
B
B
B
B
D
D
C
D
A
D
A
B
A
B
B
D
D
B
Yeah,
okay,
now
we're
going
to
talk
about
some
papers
on
organogenesis
and
organoids,
so
we
talked
a
little
bit
in
the
tensegrity
discussion
about
using
organoids
as
a
way
to
look
at
shape
changes
in
tensegrity,
but
this
is
more
about
organoids
in
general
and
so
there's
some
interesting
papers
that
have
come
out
in
the
past
week
or
so
on.
Organoids.
B
So
the
first
article
is
a
nature
paper.
Embryoid
model
completes
gastrulation
and
relation
in
organogenesis,
so
gastrulation
and
relation
is
a
is
a
milestone
in
development
of
the
development
of
embryos
and
it's
where
you
get
these
changes
from
gastrulation
to
neuralation.
So
these
are
gas
relation
intervals,
changes
in
shape
to
neuralation,
which
starts
to
involve
the
emergence
of
the
brain
and
the
nervous
system,
and
then
organogenesis
simply
means
that
you're
starting
to
form
organs.
B
B
So
this
this
is
the
abstract.
Embryonic
stem
cells
can
undergo
many
aspects
of
Amelia
embryogenesis
in
vitro,
so
you
can
have
embryonic
stem
cells
in
a
dish
or
in
a
and
some
sort
of
three-dimensional
nature
gel
and
you
can
Model
A
lot
of
aspects
of
embryogenesis,
but
having
these
cells
sort
of
in
a
culture
in
a
disembodied
culture
doesn't
really
cut
it
for
a
lot
of
things.
So
you
can't
observe
some
of
these
processes
some
of
these
processes
that
require
a
geometry
and
that's
why
we
want
to
use
organoids
to
do
this.
B
But
of
course,
the
developmental
potential
of
embryonic
stem
cells
and
culture
is
substantially
extended
by
interactions
with
extra
embryonic
stem
cells,
including
trophoblast
stem
cells,
extra
embryonic
endoderms
stem
cells
which
are
exems
and
inducible
xen
cells
or
ixcns.
So
there
are
three
different
classes
of
cell
here:
the
tropical
triple
blast:
stem
cell,
just
TS
cells,
the
extra
embryonic
endoderm
stem
cell,
which
is
xcn,
which
is
for
extra
and
then
inducible,
xcn
cells,
ixen.
So
the
little
I
is
the
inducible
aspect.
B
B
Some
sort
of
coherent
set
of
systems,
our
embryo
model
displays
head
folds
with
defined
forebrain
and
midbrain
regions
and
develops
a
beating
heart-like
structure,
a
trunk
comprising
a
neural
tube
and
somites,
a
tailbud
containing
neuromesodermal
progenitors,
a
gut
tube
and
primordial
germ
cells.
So
you
can
see
that
this,
this
embryo,
that
the
embryo
model
that
they're
building
involves
a
number
of
different
sort
of
shape.
You
know
very
shapely
structures,
so
you
have
of
course
head
Folds,
some
sort
of
Define,
forebrain
and
midbrain
regions.
So
you
have
the
this
neurogenesis
going
on.
B
You
have
not
only
the
Genesis
of
neurons
but
the
Genesis
of
the
neuronal
phenotype.
You
have
heart
leg
structure,
so
you
have
heart
cells
that
are
they're,
beating,
I
guess
so
they
have
this
Pacemaker.
And
then
you
have
a
trunk
comprised
of
a
neural,
tube
and
somite.
So
you
have
the
neural
tube
form
with
somites
a
tailbud
with
neuromusodermal
progenitor.
So
you
have
all
these
different
parts
of
the
nervous
system
that
are
going
to
be
put
together
later
or
in
this
case
just
those
parts
of
the
nervous
system
that
are
being
recapitulated.
B
So
that's
important
because
we
can
see
the
sort
of
the
origins
of
the
nervous
system
and
maybe
how
to
engineer
that
this
complete
embryo
model
develops
within
an
extra
embryonic
yolk.
Sac.
Remember
we're
talking
about
mice,
so
there's
a
yolk
Sac
with
in
a
in
a
uterus
that
initiates
blood
Island
development.
So
we
have
we're
actually
getting
some
sort
of
blood
Island
development
here
as
well,
so
we're
getting
these
systems
put
together.
So
it's
not
just
cells.
You
know
collectively
behaving
they're
actually
forming
these
tissues
and
forming
these
systems.
B
Most
importantly,
notably,
we
demonstrate
that
the
neural
lighting
embryo
model
assembled
from
PEC
6
knockout
es
cells,
aggregated
with
whale
type
TS
cells.
So
these
are
the
embryonic
stem
cells
and
then
the
terpal
blast
stem
cells
TS.
So
these
are
aggregated
together.
Actually
in
the
esls
OR
pac-6
Knockouts
pac-6
are
involved
in
eye
development
and
then
wild
type,
PS
cells
don't
have
a
knockout
they're,
just
regular
TS
cells
and
I
ixcn
cells.
So
these
are
induced.
B
So
in
pack,
six,
now,
if
you
do
some
experiments,
you
knock
out
pack,
six
in
a
normally
developing
mouse
or
in
a
mile
in
a
in
a
mouse.
That's
bio!
Not
like
this
model,
but
the
actual
Mouse.
You
can
knock
out
pack
six
and
you
get
a
phenotype.
That's
actually
you
know
a
little
bit
different
than
what
you'd
find
in
normal
development,
but
you
get
a
recapitulation
of
this
phenotype
for
The
pac6
Knockout
embryos.
So
this
is
important
that
we've
been
able
to
build.
B
B
That
goes
over
a
lot
of
this.
This
is
a
figure
here
where
they
show
what
they're
doing
here
they
have
their
different
cell
types,
they're,
putting
them
in
to
a
common
play.
A
common
culture
then
they're
starting
to
grow
these
embryo
or
these
embryo
models.
So
this
is
where
they're
starting
to
the
shape
is
starting
to
take
place.
B
So
you
have
this
fertilization
event
and
then
you
have
these
different
cell
types
and
then
this
is
where
you
get
the
inner
formation
of
the
Inner
Cell
mass
in
the
mouse
embryo,
and
then
you
get
the
different
cell
types
as
layers
up
in
this
part
of
this
embryo
and
then
around
the
edge.
So
you
can
see
the
different
cell
types
here,
labeled
by
color,
and
then
those
cell
types
then
go
to
the
when
you
start
to
get
an
embryo,
so
you
start
to
get
changes
in
shape
use.
B
You
know
over
time
you
start
to
get
the
cell
types,
but
they
migrate
to
different
poles
in
the
in
the
embryo.
So
you
get
the
es
epiblast
region,
you
get
the
TSC
trifectaderm
region,
and
then
you
get
all
this
like
this.
So
this
is
the
trifecta
Derm.
It's
going
to
transform
into
this
it's
going
to
unfold
into
this
structure,
and
then
you
start
to
get
other
types
of
tissues
that
form
later
so
heart
and
the
gut
tube
and
things
like
that.
B
And
then
you
start
to
get
a
little
bit
more
formation
here
with
a
primitive
streak.
And
then
you
start
to
get
even
more
formation
here.
So
you
have
this.
You
know
the
you
have
the
trifecta
Derm
up
here,
the
Korean
and
some
of
these
in
the
oak
sac
eventually
coming
up
here.
So
this
is
this
would
be
the
trifecta
German
Corian,
and
then
this
is
the
yolk
Sac.
So
this
is
the
transformation
that
occurs
in
this
and
then
this
is
an
analysis
here
of
I
believe
this
is
some
sort
of
multi-dimension.
B
This
is
a
umap
analysis
of
the
different
cell
types,
so
you
can
see
that
they're
I
think
they're
expressing
different
genes
here
or
are
there
different
days,
so
they
they
show
the
difference
between
natural
embryos
and
these
etix
embryos,
and
so
you
can
see
that
they're
somewhat
similar
but
different.
So
they
have
this.
It's
hard
to
interpret
these
umap
patterns,
but
they
basically
show
these.
They
basically
have
the
same
clusters
with
the
same
constituents
more
or
less,
but
they're
a
little
bit
different
in
terms
of
their
where
they're
expressed.
D
B
Again,
this
is
not
like
specific
to
any
area
or
region
of
the
embryo.
This
is
just
a
umap
analysis
and
visualization,
so
you
have
all
these
different
tissue
types
and
cell
types
that
are
put
into
this
different
colors
and
you
can
see
them
in
a
clustering
in
a
in
a
certain
way.
So
it's
very
it's
similar
in
a
lot
of
ways
to
the
Natural
embryos,
but
their
differences
as
well,
and
so
this
is
an
example.
Here
too,
of
some
of
you
know
they
kind
of
break
this
down
a
little
bit
more
here
in
F.
B
This
is
the
annotated
and
combined
umap
natural
embryos,
cultured
X,
utero
and
collected
indicated
time
points
and
etix
embryoids
individually,
labeled
and
analyzed
by
tiny
PSI,
rna-seq,
so
they're
using
rna-seq
markers
to
Define
these
clusters
some
of
these
in
these
types.
So
you
can
see
that
here
so
yeah.
There
are
differences
in
in
similarities.
B
Then
we
get
into
this
part
of
the
paper
where
we
talk
about
the
they
talk
about
how
it
develops
an
anterior
brain
and
pattern,
neural,
tube
and
I'm,
referring
to
the
Embryo
Embryo
models.
So
here
you
have
some
of
the
images
here
where
they
look
at
different,
the
expression
of
different
genes.
So
this
is
EA
embryo
here,
where
you
have
socks,
one
and
brachiary
being
expressed
in
the
embryo.
This
is
I,
think
the
these
are
different
views
of
the
embryo,
so
this
part
is
expressing
socks.
B
B
B
Mainly
it's
not
really
that
much
different
in
terms
of
the
direction
of
expression
or
the
actual
expression
of
these
things.
So,
for
example,
you
see
that
there's
a
fraction
of
this
is
the
the
size
of
these
circles,
the
fraction
of
cells
in
the
group,
the
percentage
of
the
cells
in
a
group
so
that
sometimes
the
number
of
cells
changes.
Sometimes
the
level
of
expression
changes,
but
it's
always
basically
in
a
similar
Direction.
So
that's
good
and
then
finally
we
have
this
pack
6
knockout
work.
B
So
this
is
in
the
embryo
models
where
these
pack
six
Knockouts,
they
recapitulate
what
we
see
in
the
mouse
embryonic
phenotypes,
so
we
have
in
a
mouse.
We
can
knock
out
pack
six
and
we
have
a
certain
phenotype
that
results
it's
different
from
the
wild
type,
and
so
then
we
can
ask
well,
can
we
do
this
in
an
embryo
model
and
indeed
we
can-
and
this
is
showing
the
evidence
of
this-
then
they
do
some
annotation
here,
so
the
sanitation
experiment
is
actually
so.
B
This
is
a
gene
ontology
analysis,
so
this
is
where
genes
are
they.
They
have
an
ontology
system
where
they
label
the
genes
with
a
function
and
sometimes
the
function
is
Trivial,
but
sometimes
it's
somewhat
specific.
So
you
can
look
at
the
biological
processes
here.
These
go
terms
and
you
can
get
a
sense
of
what's
what
genes
are
enriched
for
what
functions.
So
you
know
these.
These
categories
are
not
necessary,
they're
imperfect,
but
they
give
you
some
sense
of.
D
B
Are
highly
enriched,
and
so
that's
that's
interesting,
because
that
suggests
that
there's
a
lot
of
motility
going
on
at
this
in
this
at
this
stage
of
development
in
the
embryo
models.
So
this
is
where
you
have
a
lot
of
these
different
genes
related
to
movement,
movement
towards
stimulus,
chemical
stimulus
orientation
of
the
cells
and
so
forth.
So
there
are
a
lot
of
things
and
then
there
are,
of
course,
these
genes
that
are
involved
in
axonogenesis
neurogenesis,
nervous
system,
development,
axon
developments.
B
We
also
have
as
a
final
figure,
the
initiation
of
gut
development,
and
so
this
is
again
like
the
eye
and
the
defined
mutant
of
the
eye.
We
have
this
gut
development,
looking
between
the
embryoid
model
or
the
embryo
model,
and
then
in
the
natural
biological
embryo,
and
so
you
can
see
that
they
show
again,
they
show
the
co-expression
of
genes
and
morphology.
So
you
can
register
the
gene
expression
with
the
morphology
and
you
see
some
differences
in
the
morphology.
B
The
gene
expression
is
generally
in
the
same
place
with
some
with
some
differences,
because
the
morphology
is
a
little
bit
different
and
then
this
part
here
where
they
show
the
same
graph
where
they
show
the
fraction
of
cells
they
show
the
different.
Actually,
in
this
case
it's
the
different
tissues
that
are
forming
and
the
different
genes
that
are
being
a
regulated
or
done
regulated.
B
So
this
is
the
z-score
again
negative,
2
would
be
down
regulated,
positive
two
would
be
a
percolated,
and
you
can
see
that
there's
a
lot
of
upregulation,
not
a
whole
lot
of
down
regulation
in
these
structures,
and
then
this.
This
shows
you
the
difference
between
the
natural
embryo,
a
embryo
and
the
embryo
model
in
terms
of
these
different
cell
types
and
then
there's
a
umap
analysis
where
you
have
these
structures
here,
they're
these
clusters-
and
they
show.
This
is
an
overlapping
cluster
analysis.
B
B
So
there's
a
lot
more
information
in
that
paper.
If
you
want
to
check
it
out,
it's
a
very
good
paper.
This
is
another
paper.
This
is
a
little
bit
different.
This
is
inferring
and
perturbing
sulfate
regulums
in
human
brain
organoids.
So
this
is
where
we're
actually
getting
selfie
regulation
and
what
they
call
regulum,
which
is
I,
guess
the
set
of
genes
that
are
involved
in
regulation,
and
so
this
is
in
humans
now.
B
B
We
also
identify
temporarily
Dynamic
and
brain
region
specific
regulatory
regions.
So
this
is
where
you
want
to
know
if
things
are
forming,
what
the
regulatory
regions
are
being
you
know
activate
which
regulatory
regions
are
being
activated
and
that
can
tell
us
something
about
maybe
brain
development.
More
generally,
so
they
actually
developed
a
framework
called
Pando
which
incorporates
multi-ohmic
data
and
predictions
of
transcription
Factor
binding
sites
to
infer
Global,
Gene
regulatory
Network
describing
organoid
development.
B
We
use
pool
genetic
perturbation
with
single
cell
transcriptome
readout
to
assess
strengths,
transcription
Factor
requirements
for
sulfate
and
state
regulation
and
organoids.
So
this
is
something
where
there
are
genes
for
organoids,
it's
where
what
genes
are
being
expressed
and
how
are
they
being
regulated?
And
how
does
that
create
a
stable
organoid?
And
it
may
not
be
that
they're,
you
know
created.
You
know
that
they
respond
because
they're
organoids
are
not.
It's
just
that
they
respond
to
that
environment
that
the
cells
find
themselves
in.
B
D
B
So
we
know
this
from
a
lot
of
cellular
reprogramming
literature
where
people
are
obsessed
with
figuring
out
how
to
get
the
thing
to
to
differentiate
and
then
stay
stable
after
it
differentiates
because
you
don't
want
to
differentiating
into
like
a
cancer
cell
or
some
other
type
of
cell.
You
want
it
to
remain
your
target
cell.
So
this
is
something
that
we
can
learn
from
here
as
well.
We
show
that
the
transcription
Factor
GL
gli3
is
required
for
cortical
fate,
establishment
in
humans,
recapitulating
previous
research
performed
in
mammalian
model
systems.
B
We
measure
transcriptomic
and
chromatin
accessibility
in
normal
or
GOI
3
perturbed
cells.
So
this
is
again
where
we're
doing
not
a
knockout,
but
we're
perturbing
that
factor
in
this
in
these
cells.
So
we
have
a
wild
type
and
we
have
this
not
really
a
mutant,
but
we're
perturbing
that
gene
expression
in
that
Gene
and
identify
two
distinct
geoi3
regulums
that
are
Central
to
telencephalic
fate
decision.
So
this
is
the
telencephalic
region
of
the
brain.
B
So
this
is
an
early
one
of
the
lobes
of
the
brain,
where
you
know
I
think
it's
the
part
where
you
have
the
that
later
you'll
have
neocortex
emerge
from
so
this
is.
This
is
something
that,
where
you
have
a
regular
one,
which
is
just
a
series
of
regulatory
regions
and
and
other
things
that
coordinate
this,
so
these
tones
of
allocate
decisions
involve
a
number
of
things.
B
So
this
is
where
you
have
a
diversification
of
regions
in
the
brain,
so
this
is
where
it's
controlled,
or
these
ganglionic
Eminence,
and
how
exactly
what
that
means,
but
they
basically
it's
the
control
of
patterning
and
the
control
of
regional
diversification.
But
this
is
again
is
where
you
get
this
happening.
Sort
of
early
on
when
things
are
taking
shape
and
then
later
in
development,
where
things
are
sort
of
specializing
together,
we
provide
a
framework
for
how
human
model
systems
and
single
cell
Technologies
can
be
leveraged
to
reconstruct
human
developmental
biology.
B
B
They
point
out
here
is
that
direct
comparisons
between
organoids
and
primary
counterparts
in
Mouse
and
human
have
Quantified
a
notably
a
notable
similarity
between
the
neural
progenitor
and
neurotranscriptome
profiles.
So
this
is
a
neuronal.
Transcriptome,
of
course,
is
where
you
have
the
brain,
that's
in
place
and
there's
a
mature
brain,
and
there
are
a
lot
of
there's
a
lot
of
transcription
going
on
with
respect
to
different
functions
and
maintaining
the
brain.
Neural.
Progenitors,
of
course,
have
this
sort
of
similar
pattern
of
transcriptional
activity.
B
B
Brain
organized
have
been
used
to
successfully
model
micro
encephaly,
which
is
where
you
get
a
very
small
brain,
periventricular,
heterotopia,
autism
and
other
neurodevelopmental
disorders
that
may
have
differential
effects
in
the
various
human
brain
regions.
So
the
emergence
of
brain
regions.
You
expect
you
know,
changes
in
the
transcriptome,
but
in
brain
organoids
we've
been
able
to
model
these
different
disorders.
So
that
means
that
the
changes
that
occur
in
the
transcript
transcription
with
respect
to
these
different
disorders
also
can
be
recapitulated
in
the
in
vitro
models.
B
But
we
don't
understand
the
gene
regulatory
networks
that
coordinate
early
human
brain
development
and
normal
and
perturbed
conditions.
So
we
don't
really
understand
the
networks
we
just
understand,
maybe
the
profile,
which
means
you
just
look
at
a
bunch
of
genes.
You
look
at
how
they're
upregulated
or
non-regulated
and
then
that's
a
that's
like
a
visualization
like
we
saw
with
the
with
the
TC
examples.
B
So
this
is
a
difference,
because
the
regulatory
networks
are
more
predictive
and
they
tell
us
things
about
how
genes
work
together,
but
we
don't
know
that
we're
trying
to
figure
that
out.
Research
and
model
systems
has
identified
core
signaling
factors
in
gene
regulatory
programs
that
orchestrate
brain
region
formation
and
vertebrates.
So
we
saw
a
little
bit
about
that
in
the
last
paper.
Initially
extrinsic
signals
establish
an
anterior
posterior
axis,
so
this
is
again
like
this,
like
in
C
elegans,
where
you
have
the
head
to
tail
axis,
where
you
get
this.
B
Things
are
laid
out
in
that
direction.
The
triggers
additional
locate,
localized
gradients,
Downstream
to
segment
the
neural
tube
into
distinct
brain
regions.
So
you
start
with
a
neural
tube
it
elongates,
and
then
you
get
these
brain
regions
that
show
up
along
there,
sort
of
a
segmentation
process,
combinatorial
activities
of
morphogens,
including
shhh,
Lintz,
bmps,
fgfs
and
notch
or
regulans,
and
our
spondons
converted
on
transcription
factors.
So
the
transcription
factors
are
things
that
regulate
gene
expression.
B
You
get
these
other
orphogens
that
work
combinatorially
to
execute
regionalization,
so
you
combine
the
morphogens
with
transcription
factors
and
you
get
different
Regional
differentiation.
Much
of
what
is
known
about
these
Pathways
and
regulating
brain
morphogenesis
has
been
examined
in
non-human
model
systems
and
it
remains
unclear
how
human
brain
development
has
diverged
from
our
Mobility
ancestors.
So
these
are
things.
Maybe
we
can
study
in
a
model
such
as
this.
B
B
You
know
you
have
this
maturity
over
different
days.
You
have
the
embryoid
bodies
and
then
you
get
this
organoid
at
day61,
so
it
takes
61
days
to
grow
these
things
from
the
initial
seeding
of
the
cells.
You
get
these
embryoid
bodies
and
then
you
get
these
other
structures
and
then
the
organoids
they're,
just
showing
some
of
the
examples
of
what
people
do
with
these
different.
This
is
a
umap
visualization
here
of
the
different
cells,
and
then
this
is
abstracting
this
into
a
graph
I'm.
B
Not
really
sure
a
lot
of
the
details
of
this
figure,
but,
let's
see
per
a
of
course,
is
where
you
have
a
schematic
of
the
experimental
design,
a
new
map
embedding
of
integrated
multiomic
meta
cells.
They
just
show
the
different
cell
types
that
emerge
from
this
B
is
examples
of
loci,
with
differential
accessibility
during
organoid
development
or
temporary
potency.
B
B
Here
this
is
showing,
where
you
get
these
different
genes
and
how
these
loci
can
be
accessed
during
organoid
development,
so
sometimes
these
loci
that
are
on
these
critical
genes
that
are
necessary
for
so
this
is
power,
5
F1,
which
is
important
in
Mouse
development
and
fgf8,
which
is
also
important
in
Mouse
development.
This
is.
D
B
And
dcx-
and
it
just
shows
along
this
along
these
loci-
you
know
what
what's
sort
of
the
accessibility,
so
the
chromatin
will
open
up
and
allow
accessibility
to
these
genes
at
different
stages
in
in
the
development
of
these
organoids.
So
from
the
pses
to
the
ectoderm
to
the
epiblasts
to
the
NPCs,
the
neurons,
you
have
different
levels
of
accessibility.
B
C
is
schematic
of
the
branch
inference
strategy,
so
they're
doing
some
inference
of
these
clusters
and
they're
actually
building
I.
Guess
lineage
trees
out
of
these
the
the
graph
abstraction.
So
it's
then
they
assign
them,
they
build
a
graph
abstraction
and
then
they
assign
them
to
branches.
So
you
can
see
here
where
you
have
these
branches.
This
is
differentiation
in
the
telencephalon,
the
different
types
of
cells
here,
either
non-tone
cephalon
dorsal
tones
of
one
ventral
to
one
cf1
and
so
forth.
B
B
B
And
so
this
is
a
regulatory
Network,
where
they're
using
the
page
rank
algorithm
to
find
the
pathways
through
this
network
and
how
it
up
regulates
and
down
regulates
with
respect
to
different
parts
of
the
brain.
So
this
is
again.
This
is
where
they
can
take
this
network
and
break
it
up
into
different
functional
regions
with
respect
to
different
cell
types
and
and
things
like
that,
and
then
they
show
this
Global
grn,
where
you
have
these
different
genes
connected.
B
B
So
again,
this
is
where
you
have
this
you're
breaking
this
down
into
the
different
genes:
you're
perturbing,
different
transcription
factors
and
you're
looking
for
critical
Regulators
things
that
keep
them
the
cells
stable,
and
so
the
part
of
this
analysis
is
based
on
that
and
then
this
final
figure
or
final
figure
we're
going
to
go
over
here.
This
is
a
single
cell
multi-ohm
view
of
gli3.
Loss
of
function
reveals
distinct,
regulums
and
effectors
of
dorsal
ventral
tone
and
stuff
one
at
specification.