►
From YouTube: DevoWorm #25: GSoC project updates, light exposure effects in embryos, push-pull pattern formation
Description
GSoC Coding period (weeks 5 and 6): Updates on Digital Microspheres and D-GNNs. Discussions of papers on the effects of light exposure in Axolotl and chick embryos, and the physics of pattern formation (push-pull dynamics, liquid crystals) in chick and starfish embryo models. Attendees: Susan Crawford-Young, Harikrishna Pillai, Karan Lohaan, and Bradly Alicea
A
B
A
B
I
got
confused
as
to
what
they
were
talking
about
because
they
had
encapsuled
celia
and
they
were
turning
and
rotating
and
there
that
was
that
broke
the
left,
right
symmetry
and
then
I'm
saying
well
so
what
it's
turning
inside
this
capsule?
How
does
that
affect
the
outside
of
it
like
that?
They
hadn't
connected
all
the
thoughts.
I
guess.
A
B
A
A
A
A
C
Should
should
I
start
yeah
yeah,
okay,
hi
hi,
bradley
hi,
hi
yeah
this
week
from
18
to
twenty-fourth.
I
am
to
try
out
different
projection
methods,
so
that
is
going
on
the
ways
previously
week
was,
I
think,
mostly
covered.
You
know.
In
the
week
four
update
where
you
know
I
was
trying
out
different
angles
for
that
for
transposing
the
contours.
C
You
know
on
a
fixed
axis,
so
the
angles
projection
methods
are
somewhat
similar,
like
they're
not
done
much
changes
like
to
the
way
of
you
know
finding
the
angle
for
the
contour,
and
you
know
how
how
much
I
should
rotate
it
before
putting
it
on
the
axis
for
the
ethereum
model,
so
that
is
there.
Otherwise,
the
this
week's
thing
you
know
coming
up
with
different
projection
methods
for
generating,
for
you
know,
projecting
it
onto
the
model.
C
C
C
Yeah,
so
this
is
the
the
one
way
you
know
I'll
have
to
take
the
cropped
out
image
of
the
embryo
and
I'll
be
using
that
to
generate
a
projection
image
you
know
in
which
I'll
transpose
it
onto
an
equal,
rectangular,
projection
image
format.
You
know
where
it
will
kind
of
be
like
like
when,
when
we
look
at
the
earth,
you
know
when
we
look
at
the
earth
on
a
2d
map.
C
So
this
is
where
you
know
I
was
supposed
to
try
out
till
I
think
july
24th
different
ways
of
going
about
this
projection
method
part.
So
I'm
still
looking
forward
to
looking
and
still
keep
looking.
You
know
for
different
methods
for
doing
that
otherwise
or
the
equal
rectangular
projection
formula.
You
know
the
transformation
formula
for
doing
that
is
the
one
that
I'll
be
applying.
A
C
C
You
know
approximation
at
each
step
rather
than
you
know,
trying
out
the
different
thing.
I
think
that
would
give
give
me,
I
think,
a
better
result
because,
like
with
the
previous
set,
you
know
like,
I
was
generating
outlines
on
my
own.
So
initially
I
just
had
eight
outlines
so
using
averages.
You
know
I'm
generating
more
outlines
on
its
own,
so
I
think
I'll
have
to
you
know
another
thing
that
I'm
not,
I
think,
included
right
now
is
generating
a
metric.
C
C
C
C
C
C
A
So
I
don't,
I
don't
think
jia,
hyung
or
wataru
can
join
us
today,
but
we've
been
communicating
in
slack
about
some
of
their
work
and
they've
been
working
on
in
their
own
repo
on
a
lot
of
the
stuff.
So
I
don't
know
if
I
can
really
explain
this
well,
because
I'm
not
like
in
the
repo-
it's
not
actually
my
repo.
So
let
me
go
into
their
location
here.
I
think
it's
this
one
yeah,
he
sent
me
a
link
to
oh.
This
is
a
different
one.
A
Okay,
so
they
have
a
repository
here.
They
called
devo
graph.
Here
it
is
so
this
is
devo
graph.
This
is
where
jihang
and
wataru
and
they
have
someone
else
working
with
them
on
it.
It's
not
a
gsox
student
but
they're
going
to
work
on
it
with
them,
and
this
is
so
basically
they're
pushing
to
this
repo
and
then
they'll.
Push
to
the
divo
learn
repo,
so
they're
going
to
incorporate
this
into
divalern,
and
this
is
going
to
be
like
the
graph
neural
networks
component
of
it.
A
A
We
talked
about
that
last
week
and
some
of
the
things
that
need
to
be
like
in
place
for
that
some
of
the
checks
that
need
to
happen
some
of
the
quality
control,
but
this
is,
I
don't
know
if
I
can
render
this
notebook
in
here,
but
this
is
stage
two
and
so
they've.
Okay,
so
good
they've
prepared
a
notebook
here,
they're
doing
experiments
in
stage
two,
so
the
second
stage
is
to
convert
a
csv
file
into
a
temporal
and
directed
graph.
So
this
is.
A
They
gave
an
update
to
me
a
couple
weeks
ago
on
this
now,
and
this
was
part
of
this
idea
of
taking
the
image
data
segmenting
it
getting
numbers
out
of
it
and
then
converting
it
into
these
temporal
and
directed
graphs.
So
basically,
the
graph
structures
you
use
the
centroids
of
each
cell
and
you
build
these
graph
embeddings
or
these
these
graph
representations
and
you
use
it
well.
A
A
A
For
this
example,
but
for
a
segmented
example,
you
would
basically
have
the
xyz
coordinates.
You
might
have
the
cell
volume,
you
might
have
definitely
the
cell
id.
So
you
know
that'll
all
be
sort
of
part
of
that
input
and
let's
go
back
here
all
right.
So
this
is
back
to
the
notebook
here
and
then
they're
kind
of
working
through
then
they
pre-process
the
sample
data.
So
they
have
this.
That
is
formatted
here.
A
We
also
provide
an
api
that
connects
mother
and
daughter
cells
in
successive
temporal
graphs
and
generates
a
directed
graph.
So
these
are
things
that
I
don't
know
how
far
along
these
are
in
terms
of
implementation.
But
this
is
the
idea
and
then
finally,
okay,
so
then
this
is
actually
jia.
Hong's
part,
I
think
so
it
is
a
lot
of
it
is
sort
of
this
first
draft.
A
You
know
so
he's
got
a
lot
of
this
laid
out.
Then.
Finally,
the
visualization
and
3d
plots.
So
this
is
a
three-dimensional
plot
where
the
is
tracking
the
path
of
the
cells.
So
the
idea
is
that
they
migrate
in
in
a
you
know,
you
capture
the
data
for
a
single
cell.
It's
going
to
change
its
position
or
migrate
a
bit
and
then
it
divides,
and
then
you
have
daughter
cells
and
they
migrate
and
they
divide,
and
so
these
are
the
cell
tracts
for
that
yeah.
A
So
this
is
the
tracking
the
nuclei
over
time.
It
doesn't
really
say
which
cells
these
are.
But
these
are
mother
and
daughter
cells.
So
I
think
they're
actually
like
I,
I
didn't
really
know
you'd-
have
to
explain
this
to
me
a
little
bit
more,
but
there
you
basically
have
the
cells
that
move.
You
migrate,
the
position
migrates
and
then
so
it
may
not
actually
be
the
cell.
It
may
be
the
the
nucleus
or
the
the
centroid
of
the
cell.
A
We
call
it
the
centroid
because
we
don't
want
to
make
any
assumptions
about
the
nucleus,
because
it's
not
I
mean
it's
a
marker,
that's
expressed
in
the
nucleus,
but
we
don't
know
exactly.
You
know
it's
not
like
the
center
of
the
nucleus.
So
but
anyways
it's
a
good
proxy.
I
think
for
movement
in
in
the
cell
and
and
positioning.
A
A
And
then
this
is
another.
Oh,
this
is
an
example
of
where
you
have
the
mother
cells
in
the
daughter
cells.
So
these
green
dotted
lines
go
to
the
daughter
cells
from
the
mother
cell.
So
that's
what
stage
two
looks
like
stage:
one
is
the
data
and
then
I
think
there's
maybe,
and
then
this
is
stage
three.
A
A
C
A
C
A
Yeah
yeah
and
the
embeddings
you
know
because
we're
integrating
an
individual
evil
learn.
You
know
we're
trying
to
make
it
so
that
people
can
use
it
concurrently
with
segmentation.
So
we
have
these
pre-trained
models
that
you
can
use
to
segment
your
images
and
then
you
plug
it
into
this,
and
you
can
create
graph
embeddings
and
then
the
graph
embeddings
can
move
on.
A
You
know
you
can
use
those
in
different
ways,
so
you
can
use
them
to
you
know
as
data
analysis
things,
you
can
actually
generate
a
number
of
candidate
embeddings
and
choosing
the
best
one.
You
know
for
different
criterion,
so
we
don't
really
know
like
we
don't.
We
haven't
settled
on
any
one
criterion
because
you
might
want
to
create
a
graph
with
different
criterion.
A
You
might
want
to
create
a
graph,
for
example,
of
like
a
distance
in
the
embryo.
You
might
want
to
create
one
based
on
some
signaling
molecule.
You
have
a
hypothesis
about,
so
that
would
be
like
the
advantage
of
having
a
lot
of
different
embeddings
generated,
and
then
you
know
you
can
use
it
in
other
models
as
well.
A
Okay,
all
right
yeah,
that's
good
thanks
for
the
updates,
I'm
glad
that
everyone's
coming
along
and
we
have
the
first
evaluations
coming
up
soon,
but
those
are,
I
think,
going
to
be
pretty
standard.
I
don't
think
there
shouldn't
be
any
problem
with
that,
so
they'll
be
sending
along
like
some
feedback
for
you
to
fill
out
in
like
a
week
or
two
I
have
to
fill
as
a
mentor.
B
B
I
have
one
one
glass
in
my
glasses
and
I
see
blurred
hair
it's
more
of
a
distance
thing,
so
I
can
try
to
put
stripes
on
it
now
that
it's
sort
of
on
the
way
to
being
healed.
But
it's
I
don't
have
good
vision
at
the
moment,
yeah
if
you're
wondering
where
the
strikes
went,
they're
being
delayed
anyway,
yeah
my
eyes
good.
A
A
So
some
of
these
behaviors
are
goal
directed
some
are
not
cool,
directed
and
we'll
see
a
little
bit
more
about
this.
So
this
paper
is
from
the
journal
symmetry
and
this
article
is
entitled
early
in
late,
late,
embryonic
stimulation
modulates.
Similarly,
chick's
ability
to
filter
out
distractors,
so
chicks
from
the
domesticated
chicken
galaskalis
learn
to
run
from
a
starting
box
to
a
target
located
at
the
end
of
a
runway.
So
this
is
a
goal
directed
test
of
this
running
behavior
at
test.
A
So
these
are
what
this
is
is
where
they
have
chicks
that
have
hatched
and
now
they're
behaving
in
the
world,
but
they
were
reared
in
different
conditions,
so
the
first
set
of
conditions
involve
rearing
and
hatching
or
rearing
in
darkness
from
fertilization
hatching.
So
it's
complete
darkness
throughout
the
developmental
period,
the
embryogenetic
period,
and
then
the
the
other
conditions
are
chicks
that
are
exposed
in
different
ways
to
light.
So
this
is
for
42
hours
after
fertilization.
A
A
The
results
show
that
early
embryonic
light
stimulation
can
modulate
this
particular
behavioral
lateralization
complex,
comparably
to
the
late
application
of
it
via
a
different
route,
and
so
this
kind
of
goes
through
this.
So
it
is
now
well
known
that
environmental
light
stimulation
interplays
with
the
genetic
cascade
of
events
and
promoting
brain
specialization
into
different
classes
of
fission
birds.
And
so
there
are
some
reviews
here.
A
A
complex
chain
of
developmental
steps
leads
to
brain
lateralization
in
zebrafish,
starting
with
an
asymmetrical
expression
of
a
gene
network
that
controls
the
development
of
structural
left
right
differences
within
the
epithalamus,
including
asymmetrical,
peripheral
migration.
So
these
are
all
involved
in
light
perception
and
some
other
things
in
the
brain.
And
so
this
is
this.
Exposure
to
light
is
affecting
the
ability
of
these
chicks
to
basically
respond
to
light
later
and
and
it
affects
their
golden
behavior,
as
after
they've
hatched
as
a
secondary
consequence
and
the
transparent
eggs
of
the
zebrafish.
A
So
this
goes
back
to
this
idea
of
critical
periods,
and
so
critical
period
is
a
period
in
development
where
things
are
required
in
a
very
short
time
window.
So
sometimes
we
can
think
of
this
in
terms
of
plasticity,
or
we
can
think
of
this
in
terms
of
just
coming
some
mechanism
that
gets
turned
on
at
just
the
right
time,
and
so,
if
you
think
about
this
in
terms
of
a
developmental
trajectory,
we
have
development
developmental
time
here
and
we
have
some
trait
that
is
going
to
merge
here.
A
Where
there's
this
sort
of
discontinuity-
or
this
maybe
phase
transition
between
these
two
parts
of
development-
and
this
is
a
very
short
time
window-
we're
talking
about
so
this
is
what
we
might
call
a
critical
period.
So
this
critical
period
requires
certain
things
to
be
in
place
and
this
transition
can
occur
if
things
aren't
in
place
or
if
they
don't
have
the
right
stimulus
coming
in
so
stimulus
here
or
you
could
have
the
right
set
of
baseline
conditions.
A
That
can
affect
the
timing
of
this.
The
emergence
of
this
trait,
and
so
it
can
be
outside
of
this
critical
period
window
and
you
might
end
up
with
a
developmental
anomaly
like
this.
You
know
where
the
phase
transition
doesn't
occur
normally
and
you
get
an
abnormality
later
on.
So
this
would
be
like
post
hatch,
and
this
would
be
hatching,
and
this
would
be
the
post-hatch
phenotype,
so
the
phenotype
would
differ
based
on
what
happens
or
doesn't
happen
during
this
critical
period.
A
So
that's
a
that's
a
very
brief
overview
of
critical
periods,
but
I
think
it's
essential
to
understanding
what's
going
on
here
so
again,
this
is
a
nice
paper
that
talks
about
some
of
these
different
effects
of
rearing,
the
embryo
in
different
conditions,
either
light
or
dark,
and
then
what
that
does
later
on
in
the
embryos
development.
They
talk
a
lot
about
asymmetries
here,
which
is
interesting
because
you
know
this
is
not.
This
is
very
far
from
c
elegans.
A
This
is
in
chicks
and
they
also
talked
about
different
other
that
you
talk
about
fishes.
So
you
know
these
are
all
these
are
all
vertebrate
nervous
systems.
So
it's
a
little
bit
different.
There's
that
lateralization
aspect
that
actually
plays
quite
a
role
in
some
of
these
things.
So
this
is,
there
are
a
lot
of
interesting
experiments.
You
can
do
here,
so
one
is
here:
reversing
the
eye
exposed
to
light
by
untwisting
the
embryo's
head
and
applying
a
patch
to
the
right
eye
causes
the
pattern
of
asymmetries.
A
A
A
There
is
this
third
eye
on
the
top
of
our
heads
is
like
a
photic
patch
that
would
take
in
light
stimulus
and
in
birds
because
they
have
very
thin
skelet.
They
have
a
very
thin
skull,
light
comes
through
their
skull
and
and
is
processed
by
the
pineal
gland
and
in
our
pineal
gland
we
don't
get
a
weight
input,
but
it's
usually
used
to
regulate
our
moods,
regulate
some
of
our
photic
interactions.
A
You
know
when
you
get
depressed
in
the
wintertime,
sometimes
that's
because
there's
a
lack
of
light
and
there's
a
pathway
from
the
through
the
pineal
gland
that
affects
that.
So
it's
kind
of
an
interesting,
interesting
organ,
but
this
is
actually
the
habanero
hypothesis,
so
this
abstract
reads:
releases
of
innate
responses
are
more
effective
in
many
vertebrates
when
seen
by
the
left
eye.
Again,
we
have
this
asymmetry
between
left
and
right
eye
in
zebrafish
brachydonio
rario.
A
A
If
you
give
it
a
certain
amount
of
light
exposure,
just
post
fertilization,
you
get
this
asymmetrical
effect
later
on.
We
show
here
using
response
to
a
model
predator
in
day,
seven
that
absence
of
light
on
day.
One
pf,
which
is
this
is
pf
as
post
fertilization,
so
there's
actually
a
response
to
a
model
predator
on
day,
seven
of
post
fertilization.
A
Absence
of
light
on
day,
one
post
fertilization
alone
causes
high
responsiveness
to
shift
from
left
eye
to
right
eye
into
intensify
absence
on
day
two
post
fertilization
or
dave.
Three
post
fertilization
produces
lesser
shifts,
whereas
absence
on
all
three
days
reduces
responsiveness
without
any
shift.
So
you
can
see
that
there
are
different
shifts
and
different
lateralities
and
then,
if
you
don't
have
any
sort
of
light
stimulus
at
all,
then
there's
no
enhanced
responsiveness
at
all.
A
A
known
disturbance
of
a
gene
nrp1a
and
its
expression
causes
rerouting
of
the
outflow
of
the
left,
lateral
hepanula
on
the
weigh
station
of
the
right,
medial,
habitual,
providing
a
possible
explanation
of
shifts.
So
this
is
a
sort
of
an
information
processing
thing
that
intersects
with
gene
expression.
So
if
you
disturb
gene
expression
of
a
certain
gene,
there's
a
different
path,
that's
taken
by
the
information
from
one
part
of
the
habit
to
another
and
it
causes
this
asymmetry
to
occur,
and
so
we
don't
really
know
what
maybe
the
selective
pressure
for
that
would
be.
A
Maybe
there
is
a
selective
pressure,
maybe
it's
this
neutral
change
that
doesn't
really
you
know
it
just
happens
to
be
there
and
you
get
it
triggered
with
light
exposure.
We
don't
really
know
variation
and
exposure
of
eggs
to
light
is
likely
to
produce
inter-individual
variation
in
the
field.
So
again,
this
is
this
is
variable
by
individual,
and
so
this
paper
kind
of
goes
through
some
of
the
evidence
from
chick,
but
then
also
integrates
it
with
zebrafish.
A
So
again,
this
is
a
very
interesting
paper
and
you
see
some
of
this
evidence,
perhaps
of
critical
periods
here
as
well,
because
you
need
to
do
this
during
a
certain
time
in
development,
a
very
narrow
window
and
then-
and
these
things
have
to
be
in
place
or
else
you
can't
get
the
effect.
So
if
you
expose
it
to
light
later
in
development,
many
days
post
fertilization,
you
won't
get
the
same
effect.
A
The
final
paper
here
of
this
group
is
early
late.
Embryonic
stimulation
suggests
a
second
route,
lyogene
activation,
the
cerebral
lateralization
invertebrates.
So
this
is
all
all
across
vertebrates
here
include
zebra,
fishes
and
chicks,
and
probably
some
other
organisms
as
well.
I
think
they're
actually
looking
specifically
at
chicks,
but
this
this
kind
of
ties
together,
the
other
two
papers.
This
is
in
the
journal.
A
So
again
we
have
these
asymmetries
that
we
saw
in
the
zebrafish
and
actually
in
the
chick
and
the
other
experiments
replicated
here
in
another
point
in
development.
However,
some
brain
asymmetries
develop
an
absence
of
embryonic
light
stimulation,
so
brain
asymmetries
aren't
just
due
to
light
stimulation.
There
are
other
reasons
for
them,
but
you
know
this
light
stimulation.
A
Furthermore,
early
light
action
affects
lateralization,
the
transparent
zebrafish
embryo
before
their
visual
system
is
functional,
so
this
is
important
because
we've
talked-
I
talked
at
the
beginning
of
this
section
about
how
muscle
twitching
actually
behaves
like
that
too
muscle
twitching
occurs
quite
commonly
before
muscle
is
connected
to
the
connective
or
the
nervous
system.
The
central
nervous
system-
and
so
the
same
happens
same-
is
also
true
that
some
of
these
lateralizations
occur
before
the
visual
system
is
functional.
So
this
is
visual
system
independent.
A
It's
just
something
that
happens
in
the
embryo
and
then
affects
the
visual
system
later.
Here
we
investigated
whether
another
pathway
intervenes
in
establishing
brain
specialization.
We
expose
chick's
embryos
to
white
before
their
visual
system
was
formed.
We
observed
that
such
early
stimulation
modulates
cerebral
lateralization
in
a
comparable
vein
of
late
light
stimulation
on
active
retinal
cells.
So
now
they're,
looking
at
just
stimulating
the
retinal
cells
without
like
worrying
about
the
visual
pathways
being
in
place
or
not.
A
Our
results
show
that,
in
a
higher
vertebrate
brain
a
second
route
likely
affecting
the
genetic
expression
of
photosensitive
regions
acts
before
the
development
of
a
functional
visual
system,
so
in
in
what
they
call
higher
vertebrates.
There
actually
is
a
second
round
of
light
stimulation,
and
this
is
actually
involves
the
the
pineal
gland
as
well
neil
route
of
of
sensing
light.
A
So
there
actually
suggests
that
there
are
multiple
sensitive
periods
or
multiple
critical
periods,
and
we
talked
about
critical
periods,
and
I
just
showed
you
an
example
of
one
potential
critical
period,
but
you
can
have
critical
periods
throughout
development
and
they
can
build
upon
one
another.
So
you
can
have
one
critical
period
that
has
to
be.
You
know
that
allows
for
maybe
some
change
to
occur
or
some
stimulation
and
then
another
critical
period
later
and
then
another
one
later
and
they
compound
upon
one
another.
A
A
A
A
A
And
so
there
are
a
number
of
things
going
on
here
that
I
want
to
kind
of
go
over,
and
so
this
first
paper
is
from
quantum
magazine
it's
an
article
that
basically
sets
this
up
as
an
alternative
to
the
reaction
to
fusion
models
of
turing.
So
embryo
cells
set
patterns
for
growth
by
pushing
and
pulling
so
patterns
that
guide,
the
development
of
feathers
and
other
features
can
be
set
by
mechanical
forces
in
the
embryo,
not
just
by
gradients
of
chemicals,
and
so
the
reaction.
A
Diffusion
model
of
turing
really
relies
on
chemical
gradients
that
rely
on
differences
in
chemical
gradients
and
other
types
of
chemical
interactions
to
form
patterns.
But
in
this
case,
what
they're
arguing
is
that
and
you'll
see
that
there's
a
group
of
people
working
on
this,
at
least
one
group
where
mechanical
forces
are
really
driving
a
lot
of
this
pattern.
A
A
A
So,
as
you
know,
the
cells
in
the
tissues
push
and
pull
on
each
other
and
on
the
support
of
protein
scaffolding
or
the
extracellular
matrix
to
which
they
are
linked,
and
so
some
researchers
have
suspected
that
these
forces,
coupled
with
changes
in
pressure
and
rigidity
of
the
cells,
might
direct
the
formation
of
complicated
patterns.
We've
talked
about
this
before,
where
the
cells
you
know,
cells
can
change
their
shape
and
they
can
do
a
lot
of
interesting
things,
especially
collectively
in
this
case,
we're
actually
suggesting
that
they
form
they're.
A
The
cells
pulled
in
collagen
fibers
in
the
extracellular
matrix
as
they
assembled
around
themselves
and
then
over
48
hours,
the
fibers
rotated
bunched
together
and
then
pushed
each
other
apart,
forming
bunches
of
cells
that
would
become
feather
follicles,
so
they're
basically
forming
these
clusters
from
just
basically
a
a
distribution.
You
know
random
distribution
of
fibers
and
they're
forming
these
patterns,
so
this
is
a
lot
like
turn
pattern
formation.
A
Where
you
get
these,
you
know
sort
of
this
homogeneous
gradient
or
this
randomized
gradient,
and
then
you
end
up
with
these
patterns
stripes
or
dot
or
spots
or
whatever.
In
this
case
they
would
basically
be
spots,
so
you
can
see
in
the
image
here
that
form
and
form
the
basis
for
these
feather
follicles.
A
Okay.
So
this
is
this.
Is
such
a
clean,
simple,
experimental
setup
where
you
can
see
a
beautiful
pattern
come
out
and
quantitatively
control
it,
and
so
you
can
do
a
lot
of
things.
This
is
a
very
quantitatively
oriented
experiment.
So
you
know
there's
there
are
a
lot
of
parameters
that
you
can
adjust
and
look
at
like
how
this
works.
A
A
We
find
that
early
in
it
in
gut
development,
proliferating,
progenitors
expressing
isc
markers
which
are
intestinal
stem
cell
markers,
so
they
have
molecular
markers
that
they
can
identify
progenitor
cells
as
how
is
possessing
are
evenly
distributed
throughout
the
epithelium
in
both
check
and
mouse.
So
they
have
an
even
distribution
across
that
epithelial
layer.
However,
as
the
vli
form,
the
putative
stem
cells
become
restricted
to
the
base
of
the
voi.
A
The
shift
in
the
localization
is
driven
by
mechanically
influenced
reciprocal
signaling
between
the
epithelium
and
the
underlying
mesenchyme.
So,
as
we
saw
in
the
graphical
abstract,
we
can
see
these
two
layers
and
there's
reciprocal
signaling.
You
can
see
the
arrows
here
between
wind
and
shhh
and
bmp,
and
so
there's
the
signaling
in
both
positive
and
negative
signaling.
So
and
then
we
have
buckling
forces
physically,
distort
the
shape
of
the
morphogenetic
morphogenic
field,
causing
local
maxima
of
epithelial
signals,
in
particular
shhh
at
the
tip
of
each
villus.
A
This
induces
a
suite
of
high
threshold
response
genes
in
the
underlying
mesenchyme
to
form
a
signaling
center
called
the
villus
cluster.
Those
cluster
signals,
notably
bmp4
feedback
on
the
overlying
epithelium
to
ultimately
restrict
the
stem
cells
to
the
base
of
each
phyllis.
So
it's
this
combination
of
signaling
and
then
buckling
forces
through
sort
of
these.
These
buckling
sort
of
these
bucklings
of
the
initial
bucklings
of
the
tissue
that
drive
the
formation
of
these
voi.
So
it's
basically
driving
the
positive
feedback.
A
I
suppose
that
drives
this
process
forward
and
refines
these
voi
from
buckner
buckles
to
these
for
foreign
fi,
and
it
goes
across
the
surface
of
the
intestine,
so
they
show
some.
They
show
some
images
here.
They
show
an
example
here,
where
you
have
these
buckles
that
form
ui.
Eventually
it's
kind
of
like
they
look
like
fingers
coming
out
of
the
surface
of
the
intestine.
A
And
so
that
now
I'm
going
to
go
back
to
the
original
paper,
I
said
that
I
was
not
going
to
present
that,
but
I
actually
now
that
we've
established
what
what's
going
on,
then
we
can
actually
go
to
this
paper,
which
is
the
paper
that
was
featured
in
the
in
the
popular
article,
and
so
this
paper
is
titled.
Reciprocal
cell
ecm
dynamics
generate
supercell,
supracellular,
fluidity
underlying
spontaneous
follicle
patterning.
A
A
You
have
some
symmetry
breaking
event
where
you
get
this
buckling
and
then
this
refinement
of
the
buccals-
and
this
is
the
follicle
pattern,
so
this
is
a
little
bit
different.
This
is
follicle
patterning
you
get
they're
able
to
do
this
in
culture,
then
they're
able
to
show
these
orderings.
So
as
the
cells
get
plated
on
this
on
this
two-dimensional
culture,
they
form
the
cells
form
a
ring.
Then
this
ring
becomes
ordered
over
time.
So
these
are
isotropic
cells.
These
are
just
kind
of
laid
down.
A
You
know
in
a
predictable
pattern
around
in
a
circle,
then
they
become
aligned
to
the
pattern
axis
and
they
become
ordered
as
an
active
fluid,
and
then
you
get
these
aggregates
that
form
from
this
ordering,
so
they've
become
aligned,
they
become
ordered,
and
then
they
become
periodic.
We
talked
about
this
in
the
liquid
crystal
biology
lectures,
and
this
was
about
maybe
six
months
to
a
year
ago,
where
we
talked
about
pneumatic
orderings,
and
so
this
is
very
similar.
This
is
an
active
fluid.
This
is
a
soft
material
type
approach
to
this.
A
So
you
get
the
soft
material
alignment.
These
are
collective
behaviors
among
cells
and
they
have
these
different
properties.
So
refer
back
to
that
those
lectures
or
the
dejan
a
book
on
active
matter,
soft
matter,
then
in
cells,
you
see
that
you
have
this
extracellular
matrix
here.
You
have
contractile
cells
that
align
the
extracellular
matrix.
A
So
the
highlights
of
this
article
or
an
ex
fibo
assay
reconstitutes,
the
process
of
embryonic
skin
symmetry
breaking
mesenchymal
mechanics,
are
sufficient
to
spontaneously
generate
morphological
pattern.
Contractility
dependent
cell
extracellular,
matrix
interplay,
so
the
interplay
between
the
cells
and
this
extracellular
matrix
down
at
the
bottom
here
creates
supracellular
structure.
A
So
supracellular
structure
is
the
sort
of
how
in
culture
you
can
have
this
alignment
and
then
this
clustering-
and
this
is
just
in
culture,
but
in
a
tissue
where
you
would
have
this
extracellular
matrix,
you
actually
get
these
clusters
and
they
form
with
the
extracellular
matrix
around
them.
So
you
get
these
instabilities.
You
get
these
dynamics
that
you
don't
necessarily
get
in
cell
culture,
although
you
can
demonstrate
the
pattern
formation
in
cell
culture,
theory
and
experiments
show
that
cell
ecm
layer
behaves
as
an
active,
contractile
fluid.
So
again,
this
is
the
contractile
fluid
here.
A
So
what
do
we
go
over?
The
summary
during
vertebrate,
embryogenesis
cell
collectives,
engaging
coordinated,
behavior
to
form
tissue
structures
of
increasing
complexity,
and
this
set
of
experiments
is
an
avian
skin.
So
an
avian
skin
assembling
into
follicles
depends
on
intrinsic
mechanical
forces
of
the
dermis,
but
how
cell
mechanics
initiate
pattern?
A
You
know
this
pattern
formation
if
it
can
be
done
in
an
dissociated
context,
but
also
in
a
context
of
a
tissue,
so
we
find
that
contractile
cells
physically,
rearrange
the
extracellular
matrix
and
that
this
matrix
rearrangement
further
aligns
cells.
So
the
extracellular
matrix
reinforces
a
lot
of
these
pattern
formation
trends.
A
This
exchange
transfer
is
a
mechanically
unlinked
collective
of
dermal
cells
into
a
continuum
with
coherent
long-range
order.
So
this
is
the
cinematic
ordering
where
it's
long
range
and
it
you
know
it
isn't.
Just
neighboring
cells-
it's
you
know
the
50th
cell
down
the
line.
It's
also
ordered
in
the
same
way.
So
you
have
to
you
know
that
sort
of
we
don't
want
to
call
it
global
ordering,
but
that
sort
of
long
range
ordering
is
essential
for
this
sort
of
pattern:
formation,
combining
theory
with
experiment.
A
Our
study
illustrates
a
role
for
mesenchymal
dynamics
and
generating
cell
level
ordering
in
a
tissue
pattern
level
patterning
through
a
fluid
instability
processes
that
may
be
at
play
across
morphological,
symmetry
breaking
contexts,
and
we
talked
a
lot
about
symmetry
breaking
and
and
instabilities,
and
things
like
that
in
our
group
as
well.
So
you
know
we
go
back
to
some
of
the
lectures
on
that.
If
you
want
to
know
more,
so
I
don't
know
if
there
are
any
nice
figures
in
here.
A
This
just
sort
of
shows
the
alignment
in
their
experiments
and
then
pattern
formation.
So
it's
not
just
enough
to
have
pattern
formation.
You
have
this
alignment
of
cells
that
has
to
initiate
pattern
formation,
so
you
can
see
that
here
and
then
you
know
this
is
the
protocol
for
what
they're
doing
it's
worth.
Noting,
though,
that
you
have
some
of
these
things.
So
let's,
let's
take
this
figure
here,
you
have
it
figure
1a.
This
is
a
bright
field
image
of
day,
6.5
embryo
back
skin.
A
So
this
is
from
the
embryo,
and
this
is
at
day
6.5
and
then
f
is
a
bright
field
image
at
day,
7
embryo
back
skin.
So
this
is
a
half
a
day
later
in
development,
e7
and
so
maximum,
and
so
this
just
kind
of
shows
the
patterning
as
it's
happening,
so
it
it
aligns
first
in
that
e6
e6.5
window
and
then
e7.
A
Okay,
so
that's
that's!
I
think
all
I'm
going
to
talk
about
with
that
paper.
So
we
have
two
examples
of
that
now.
I
want
to
talk
more
directly
about
living
crystals
and
some
of
the
the
liquid
crystal
biology
stuff
that
I
talked
about
earlier.
If
you
go
back
to
degenerate,
it
talks
about
soft,
active
materials
and
some
of
these
phenomena
and
crystals,
but
then
makes
the
application
also
to
some
forms
of
single
cell
biology.
And
of
course
this
has
been
worked
out
in
more
recent
past
in
biophysics
in
the
biophysics
of
literature.
A
A
Whether
persistent
crystalline
order
can
emerge
in
groups
of
autonomously
developed
multicellular
organisms
is
currently
unknown,
but
here
we
show
that
swimming
starfish
embryos
spontaneously
assemble
into
chiral
crystals
that
span
thousands
of
spinning
organisms
and
persist
for
tens
of
hours.
So
we're
looking
at
starfish
embryos
here
and
I
think
okay,
this
is
the
paper
from
peter
foster,
so
susan
earlier
mentioned
peter
foster,
and
this
is
that
work
of
that
group.
So
this
is
this-
is
swimming
starfish
embryos
and
they're
spontaneously
assembling
into
these
crystals
combining
experiments,
theory
and
simulations.
A
We
demonstrate
that
the
formation
dynamics
and
disillusion
of
these
living
crystals
are
controlled
by
the
hydrodynamic
properties
and
natural
development
of
embryos.
Remarkably,
living
chiral
crystals
exhibit
self-sustained
thyroid
oscillations,
as
well
as
unconventional
deformation
responses.
Behaviors
recently
predicted
fraud,
elastic
materials.
A
Our
results
provide
direct
experimental
evidence
for
how
non-reciprocal
interactions
between
autonomous
multicellular
components
may
facilitate
non-equilibrium
phases
of
chiral
active
matter.
So
this
is,
you
know
again
with
this
is
now
with
a
very
different
type
of
organism:
starfish
embryos
and
we
have
symmetry
breaking.
A
We
have
some
of
these
active
matter
issues,
and
so
this
is
so
a
particular
interesting
class
of
non-equilibrium
symmetry
breaking
phenomena
comprises
the
active
crystallization
processes
recently
observed
in
colloidal
and
bacterial
systems,
so
colloidal
systems
are
systems
of
particles,
usually
materials
like
sand
or
other
types
of
soil
or
other
types
of
things
that
that
have
this.
You
know
collective
behavior
and
they
form
states
that
matter
unlike
conventional
passive
crystals,
which
form
on
lowering
the
temperature
and
often
require
attractive
forces.
A
A
long-standing
related,
unanswered
question
is
whether
groups
of
multicellular
organisms
self-organize
in
the
states
of
crystalline
order,
so
crystalline
order
is
the
sort
of
macrostate.
So
in
this
case
you
have
starfish
embryos,
they're
independent,
but
then
they
form
these
macrostates,
and
these
macrostates
are
the
crystals
that
you
talk
about
here.
So
they're,
not,
I
guess,
they're
living
systems,
but
they're
not
like
you
know.
The
macro
state
isn't
necessarily
a
living
system,
but
the
the
constituent
parts
are
so
here.
A
We
report
the
observation
of
spontaneous
crystallization
large
assemblies
of
developing
starfish
embryos,
patria
minotata,
miniata
arcsbino.
Our
experimental
observations
show
how,
over
the
course
of
their
natural
development,
thousands
of
swimming
embryos
come
together
to
form
living
chiral
crystal
structures.
They
persist
for
many
hours
and
they
have
a
video
of
this
which
I'm
not
going
to
show,
but
it's
a
supplemental
video
to
this
paper,
in
contrast
to
externally
actuated
colloidal
systems,
the
self-assembly
dynamics
and
disillusion
of
these
lccs
are
controlled
entirely
by
the
embryo's
internal
developmental
program.
A
So
this
is
where
the
embryos
at
a
certain
life
history
a
point
in
their
developmental
history.
They
form
these
constructs
and
then
they
just
dissolve
as
quickly
as
they
formed.
So
this
goes
back
also
to
the
critical
periods,
because
you
have
these
little
periods
where
they
form
these
different
structures
once
formed.
These
lccs
exhibit
striking
collective
dynamics
consistent
with
predictions
from
recently
proposed
theory
of
audio
elasticity,
and
so
audio
elasticity
is
well
we'll
talk
about
that
in
a
little
bit.
I'd
like
to
go
through
these
images.
Actually,
so
these
are
images
of
this
crystal
formation.
A
These
are
the
embryos
here
and
they're,
forming
this
crystal
structure
and
then
they're
forming
patterns.
So
you
can
see
that
the
embryos
aggregate
and
they
form
these
crystal
structures,
and
so
they
self-organize
into
living,
chiral,
crystals
and
sequences
are
still
images
showing
crystal
assembly
and
disillusion.
A
This
is
a
is
at
two
hours
and
then
this
one's
at
five
hours,
26
hours,
35
hours
and
38
hours.
Basically
it
shows
a
sequence
where
they
start
to
build
up.
They
build
this
crystal.
You
can
see
pattern
formation
within
the
crystal
and
then
it
dissolves,
and
so
you
get
the
dissolution
at
38
hours
where
they
break
apart.
A
A
This
internal
structure
in
the
embryo,
26
30
and
44
hours
so
38
hours
when
it's
dissolving
it
looks
much
different
than
what
it
looks
like
at
12
hours
or
zero
hours,
but
actually
not
that
much
different
than
26
hours.
It's
just
the
dynamics,
change
the
flow
field,
dynamics
change
so
and
then
this
c
actually
is
embryos
assembled
in
the
crystal
perform
a
global,
collective
rotation,
so
once
they
get
into
these
crystals,
they
form
rotations
and
they
get
through
the
water
column.
A
A
Then
this
drives
this
process
and
even
if
they're,
not
necessarily
synchronized,
you
still
get
like
the
dissolution
of
crystals
as
as
some
of
the
embryos
age
out.
So
this
is
an
example
here,
embryo's,
spinning
frequency
and
then
this
is
from
zero
minutes
to
20,
to
40,
to
50,
to
60,
to
200
minutes,
and
you
can
see
that
the
in
this
case
the
crystal
is
persisting
over
this
time
period.
A
This
highly
ordered
state
process
for
several
hours
and
then
as
development
progresses,
you
get
changes
in
morphology
and
the
surrounding
flow
fields
and
that
kind
of
breaks
up
the
crystalline
order.
So
these
flow
fields
change
as
the
colony
sort
of
forms,
these
crystals
and
they
kind
of
rotate
around,
and
so
this
is
all
kind
of
almost
like
a
self-contained
system
and
it's
driving
its
own
dynamics
and
the
constituents
are
driving
the
dynamics,
they're
really
kind
of
a
fascinating
system.
A
A
So
starfish
embryos
are
inherently
chiral,
which
means
they
spin
in
a
certain
direction.
They
have
a
handedness,
so
they
have
like
a
bias
towards
one
direction
or
another.
They
spin
about
their
ap
axis,
which
is
anterior
posterior
axis
in
a
left-handed
manner,
so
they
spin
left-handed
and
they
prefer
that
that
orientation.
A
A
They
have
this
sort
of
some
sort
of
defect
in
the
colony,
and
so
this
is
what
they're,
referring
to
the
presence
of
odd
module,
which
are
a
derivative
sort
of
of
model.
Autoelasticity
raises
the
possibility
that
lccs
can
support
self-sustained,
chiral
waves
and
strain
cycles
similar
to
those
recently
predicted
on
elastic
materials.