►
From YouTube: DevoWorm (2023, #8): D-GNNs project, Second Law and Waddington landscapes, tempo of development
Description
Progress on and getting ready for Google Summer of Code. Articles on the History of the Second Law of Thermodynamics. Biological systems as open systems (and open systems vs. closed systems). Living with Waddington (epigenetic) Landscapes, The tempo of developmental time in the genotype and phenotype. Attendees: Richard Gordon, Susan Crawford-Young, Sushmanth Reddy Mereddy, Kritika Verma, Jesse Parent, and Bradly Alicea
A
Well,
Bradley,
hello:
how
are
you,
okay,
how's
everyone
doing
this
month
and
krika
and
Jesse
and
Dick
and
Susan.
C
A
Oh
Jesse
said:
can
I
get
an
invite
to
the
diva
worm
slack?
What's
the
open
arm,
slack
I,
don't
think?
Okay,
you
have
to
do
it
through
your
Joe
Pro
account.
Okay,
well,
I'll,
give
you
the
invite.
Actually
it
should
go
I'll
just
send
you
the
link
later.
A
Yeah,
so
so
any
hip
updates
from
anyone.
Anyone
wanna
anything
to
talk
about
is
doing
some
things,
I
think
more
things
with.
D
D
F
Pretty
stuck
with
creating
a
API
for
developer
on
plug
increase,
I
took
I
was
taking
help
from
one
guy
from
hubby
face
to
where
Adams
he
was
also
contributing.
Actually,
two
days
back,
he
kept
a
full
request
to
our
report.
I
imagine
it
thought
your
permission.
Actually
I
didn't
let
you
know,
but
I
mentioned
it.
Some
documentation
thing.
He
helped
me
out.
Okay
and
we
are
creating.
We
are
creating
an
API
for
development.
Like
can
I
share
my
screen
and
show
it.
F
F
I,
don't
I,
don't
have
clear
knowledge
regarding
this,
so
I
kept
it
like
somewhat
empty,
but
we
need
to
add
this
thing:
I
have
given
the
documentation
whatever
it
is
there
in
developed,
yes
and
I
want
to
create
an
API,
and
all
this
over
here
I
mean
like
halfway
there
Maybe
we'll
try
to
complete
this
work,
so
I
mean
everything
will
be
work
up
and
running,
and
and
according
to
gsock
I
mean
when
we
can.
When
can
we
submit
that
draft
proposal
actually
can
I
submit
even.
A
A
So
could
cut,
how
are
you
would
you
like
to
introduce
yourself.
E
A
So,
thank
you
sushmath
for
that.
That
was
great.
Looking
at
the
actually
just
before
we
got
into
the
meeting.
I
was
looking
at
some
of
the
things
on
hugging
face,
so
it
looks
like
it's
coming
along,
so
the
API,
that's
great
you'll
have
to
work
on
the
documentation
a
little
bit,
but
that's
yeah.
F
A
F
F
D
F
You
have
an
yeah.
F
Kind
of
regarding
Devo,
if
it's
possible,
just
a
10
minutes
meeting,
would
be
nice
yeah.
A
Yeah
be
good:
okay,
yeah,
like
some
sometime
other
than
now
like
during
the
week.
A
F
A
So
cricket
said
I
just
started
with
the
machine.
Learning
and
I'm
really
fascinated
by
bioinformatics.
I
haven't
contributed
yet,
but
I'm
really
looking
forward
to
learning
about
gsoc.
A
That's
good.
Yes,
thanks
for
attending
the
meeting,
so
you
know
please
stick
around
and
you
know.
Let
me
know
if
you
have
any
questions
or
you
find
something.
That's
really
interesting
to
you.
B
I've
discovered
a
way
to
do
my
stress
strain
graphs.
Only
I
have
to
do
my
model
in
solid
mechanics
and
not
trust
stretchers,
so
I'm
gonna
go
visit
them
at
the
conference.
Maybe
every
day
you're
gonna
feel
like
they've
been
bugged.
D
A
B
They've
got
wire
structures
and
solid
mechanics
and,
if
I
combine
the
can
combine
the
two
which
they
advertise
as
being
possible.
I'm
good,
it
is
I,
know
what
I
want
to
do.
It's
just
this
Hall
of
the
mechanics
has
a
new
level
of
difficulties
in
it
to
things,
because
you
have
to
be
able
to
geometrically
link
them
together.
The
trust
structure
you
to
say
I
want
to
draw
a
wire
from
here
to
there
and
it
does
it.
C
B
Yeah,
the
solid
mechanics,
oh
you
have
to
go
from
this
coordinate
to
that
coordinate
and
actually
draw
it.
So
anyway,
I
have
a
few
questions.
Yeah.
B
There
is
it's
coming
sort
of
a
tensegrity
modeler
out
there,
some
software
and
somebody
did
some
programming
in
Sim
link.
Okay,
that's
related
to
Matlab
content
security
structures.
B
I
know
you
want
me
to
look
those
up.
They.
D
B
B
E
Had
to
start
the
result,
Bradley
made
a
simulation
of
Random
rachenberg
vehicles,
okay
and
the
starting
result
is
that
if
we
say
that
if
we
Define
viability
by
the
ability
to
move
and
I
guess
some
a
few
other
criteria,
if
we
got
37
viability
from
random
structures,.
E
So
the
same
question
could
be
as
what
Integrity
structures.
If
we
had
a
Criterion
for
them
being
non-collaxing
okay,
then
what
fraction
of
random
structures
are
non-collection.
E
B
It
not
by
hand
yeah
well.
B
I
think
I'd
like
to
give
that
one
to
Matlab.
Actually,
okay,
so
I'll
I've
got
somebody's
thesis
that
where
they
did
their
fallen
seem
like
and
that
might
work
but
I'm
thinking
of
the
paper
that.
B
Did
some
iterations
on
tensegrity
structures
and
they
removed
the
tension,
elements
that
weren't
needed
and
the
structure
gradually
became
more
rounded
okay.
It
was
interesting.
Yeah
I
I
have
like
the
stack
of
papers
like.
B
B
B
It
might
as
well
be
yeah.
B
B
A
C
C
So,
do
you
have
any
reference
papers
well,.
A
A
We
also
have
a
website
for
the
group
which
is
here,
and
that
has
some
of
the
Publications
we've
done
in
the
group
meetings,
and
our
group
is
about
eight
years
old
so
and
that's
krika's
email.
So
you
know
you
might
find
something
of
interest.
There
basically
summarizes
what
we've
done
over
that
time.
A
So
yeah,
let's
let
me
see
if
I
can
pull
it
up.
A
Last
year,
actually,
but
this
is
the
2023
so
last
year
we
did
work
on
graph
neural
networks
and
we
kind
of
got
built
a
pipeline,
and
we
built
some
software
for
taking
using
graph
neural
networks
to
extract
graphical
information
out
of
microscopy
images.
So
you
know
microscopy
images
are,
of
course
you
know
things
that
are
allow
you
to
see
things
that
are
very
small
and
there's
a
lot
of
the
images.
Have
you
know
they're.
A
Our
images
are
basically
embryos
that
we're
training
our
graph
neural
networks
on
and
those
embryos
of
cells
that
are
have
spatial
organization.
So
that's
spatial
organization.
We
can
break
that
down
into
a
graph
structure
and
then
we
can
take
that
graph
structure
and
extract
it
using
the
GNN
and
build
embeddings
that
we
can
then
use
to
for
other
types
of
applications
like
machine
learning.
So
you
know
this
is
we?
A
Last
year
we
built
a
pipeline
to
sort
of
flush
out
how
that
would
go
from
the
microscopy
images
to
the
embeddings,
and
then
you
know
worked
on
some
of
the
software
and
but
the
problem
of
course
was
is.
That
was
a
first
pass
and
we
found
that
it
was
very
hard
to
sort
of
get
our
heads
around
what
that
should
look
like.
So
we
got
some
progress
done
last
year
and
it's
it's
sort
of
in
place,
but
it
needs
to
be
improved
and
it
needs
to
be
integrated
with
some
of
the
other
stuff.
A
We've
done
with
deep
learning.
We've
done
a
we
have
a
site
called
Divo
learn,
which
was
what
sushmath
was
working
on
through
the
hugging
face
interface,
but
there's
also
a
GitHub
repository
for
evil
learn
other.
D
A
Right
I'm
gonna
put
the
d-roller
and
Link
in
here,
so
you
can
look
at
the
GitHub
repository
and
that's
that
that's
in
the
chat
and
that
should
give
you
like
an
idea
as
to
what
the
code
looks
like
and
the
difference
so
Devo
graph
is
where
you
know:
that's
where
the
graph
neural
network
work
resides
and
then
Diva
learn
is
where
the
Deep
learning
work
resides,
and
so
this
project
this
year
is,
is
going
to
involve
sort
of
maintaining
it
moving
it
to
the
next
level
and
integrating
the
Deep
learning
with
the
gnns.
A
So
graph
neural
networks
are
a
potentially
powerful
method,
but
we
have
to
sort
of
get
our
hands
around
what
we
have.
So
what
we
want
to
do
is
develop
these
developmental
gnns
that
allow
us
to
characterize
a
growing
network
of
cells.
So
in
embryology
you
know,
cells
are
constantly
dividing
and
creating
the
structure,
and
so
we
want
to
be
able
to
take
images
of
that
process
and
extract
the
sort
of
the
graph
embeddings
that
exist
in
there.
So
there's
some
graph
structure
I
never
want
to
recover
it
and
then
build
embeddings
for
other
purposes.
A
So
we've
built
a
road
map
and
a
pipeline
for
this,
but
we
weren't
able
to
develop
full
integration
and
we
still
need
a
lot.
It
still
needs
a
lot
of
work
in
terms
of
you
know
getting
it
to
work
well
because
the
stuff
that
we
have
you
know
we
train
it
on
the
data
that
we
can
find.
And
so,
if
we
have,
you
know
there
are
a
couple
of
data
sets
that
we
have
that
we
can
train
the
models
on,
but
by
and
large
you
know
this
is
kind
of
a
hard.
A
D
A
Of
the
a
lot
of
the
data
is
not
really
optimized
for
image
segmentation,
for
example,
so
it's
hard
to
really
kind
of
use
that
effectively
but
we'll
you
know
we'll
keep
working
on
it
this
summer.
So
when
you,
if
you
propose
a
project,
you
know
your
proposal
should
focus
on
what
exactly
you
want
to
vote.
What
what
exactly
you
want
to
do
within
the
scope
of
the
gnn's
initiative?
A
So
it
could
be,
like
you
know,
taking
the
existing
code
and
improving
it
or
introducing
maybe
a
new
way
to
do
this
or
maybe
working
on
the
pipeline
working
on
the
integration
between
graph
neural
networks
and
deep
learning.
There
are
a
lot
of
ways
that
you
can
work
on
this
project.
A
I
would
suggest
looking
at
the
code
on
GitHub
and
you
know
kind
of
thinking
about,
maybe
what
you
would
want
to
do
and
then
you
know
I'll
be
available
for
feedback.
So
if
you
have
an
idea-
and
you
want
to
put
it
in
the
I.
F
D
C
A
So
yeah,
if
you
have
a
question,
you
can
ask
there
and
then,
if
you
know,
if
you
have
your
proposal,
then
you
can
pass
it.
You
can,
you
know,
send
it
to
me
and
they
can
give
you
feedback.
A
D
Account
yeah
but
I'm,
trying
to
get
up
to
date.
Again
my
schedules
worked
out
where
I
can
come
back
to
the
meetings
a
bit
more
often
and
I
want
to
kind
of
you
know:
Okay
Google,
some
real
code
a
bit
if
I
can,
but
also
there's
a
lot
of
a
lot
of
older,
like
papers
and
things
that
were
here
that
I
haven't
really
touched
on
recently
and
want
to
kind
of
pick
up
again.
D
So
no,
no
major
updates
to
me
other
than
I
hope
to
be
I
should
be
able
to
actually
physically
well
virtually
physically
meet
on
on
when
on
the
Monday
meetings.
More
often
so,
I
want
to
just
kind
of
do
that,
and
it's
good
to
see
everybody
again
and
more
in
the
future.
But
that's
all
for
me
really
good
to
see
everybody.
Oh.
A
Yeah
I
agree
thanks
for
coming
and
yeah
I
look
forward
to
you
know
if
you
want
to
be
involved
in
Google
summer
code.
Looking
forward
to
that,
definitely
you
know
we'll
try
to
do.
You
know
we'll
try
to
do
more
when
we
start
Google
summer
of
code.
You
know
we'll
do
the
community
period
and
then
we'll
do
the
coding
and
Community
period
will
be
something
you
can
help
with.
Definitely
and
then
the
coding
period
will
be
just
kind
of
like
feedback
on
things,
and
this
is
definitely
I.
A
I
get
it
let's
see,
I'll
share
my
screen.
A
A
What
do
you
mean
it's?
It
should
be
up.
E
A
E
That
would
be
fine.
Okay,
show
me
in
one
where
the
diameter
is
smaller.
E
A
C
E
A
E
Right
now,
you
know
because
of
the
Inner
Circle.
A
E
E
E
A
A
Said
I'm
on
a
treadmill
too:
it's
the
publication
cycle.
A
E
A
So
and
I
moved
to
some
things
here.
Let's
see,
I
got
a
couple
of
articles
that
are
kind
of
long
and
I'll
just
go
over
them
a
little
bit,
but
you
know
they
might
find
them
interesting,
I'll
post
them
later
or
send
them
to
you.
If
you
find
them
interesting.
D
A
First,
one
is
the
Steven
Wolfram
article
and
he
has
a
Blog
that
is
very
prolific
on
a
lot
of
things,
so
just
publish
something
on
chat
GPT,
so
you
know,
but
his
he.
D
A
A
series
on
the
second
law
of
thermodynamics-
and
you
know
this
is
like
a
three
I-
think
it's
like
he's
tried,
first
of
all,
I
think
he's
tried
to
solve
the
second
law,
so
I,
don't
know
how
effective
the
solution
is,
but
it
does.
It
does
have
some
interesting
things
to
say
about
thermodynamics
in
terms
of
its
history
and
so.
A
You
yeah,
so
this
is
kind
of
goes
over
like
the
origins
of
the
second
law,
so
the
basic
concept,
the
second
model,
is
first
formulated
in
the
1850s
and
rather
rapidly
took
on
something
close
to
its
modern
form.
A
A
This
was
pretty
early
on
in
in
the
history
of
this,
so
this
was
something
that
emerged
from
a
lot
of
things
going
on
with
the
study
of
with
mechanics
and
heat
things
like
that,
but
by
the
end
of
the
1800s,
with
the
existence
of
molecules
increasingly
firmly
established,
the
second
law
began
to
often
be
treated
as
an
almost
mathematically
proven,
necessary
law
of
physics.
A
There
were
still
mathematical
loose
ends,
as
well
as
issues
such
as
its
applicational
living
systems
and
the
systems
involving
gravity.
So
it
was.
You
know
this
is
like
the
standard
model
of
the
second
law.
Is
this
closed
system
where
you
have
a
closed
container
and
you
you
know
like
set
it
to
you,
want
to
set
it
to
equilibrium
and
then
see
how
long
you
know
entropy
takes
to
bring
it
down
to
near
Zero
Energy.
A
So
a
lot
of
the
models
of
the
second
law,
you
know
suggests
that
you
can't
you
know
you
can't
create
or
destroy
energy.
We
either
have
it
in
this
active
State
or
the
centropic
state,
and
there's
this
decrease
in
energy
or
there's
this
increase
in
entropy
over
time
that
eventually
you
get
the
system
being
mostly
entropic,
and
then
that
runs
down
into
Decay.
A
You
know,
but
that's
a
closed
system
so
usually
the
way
they
measure
it
is
just
to
close
the
system
off
to
new,
energetic
inputs
and
measure
the
Decay
they're,
of
course,
they're
open
systems
which
is
most
of
life
and
some
other.
You
know
a
lot
of
other
things,
but
that's
with
with
an
open
system.
It's
a
it's
a
different.
You
know
a
different
scenario.
So
this
is,
you
know.
A
This
is
kind
of
the
standard
thing
that
we
have
so,
but
almost
the
almost
Universal
conventional
wisdom
that
became
the
second
law
must
always
hold,
and
if
it
didn't
seem
to
in
a
particular
case,
then
that
just
must
be
because
there's
something
one
didn't
yet
understand
about
the
case.
I
don't
know
I,
guess
it's
just
that
it
applies
to
everything
and
if
it
doesn't
quite
match
up,
then
there's
something
you're,
not
understanding
about
the
second
law,
so
the
second
law
can
never
be
failed.
A
A
So
you
know
there
are
a
lot
of
application
domains
through
the
second
law
in
chemistry
and
Engineering,
for
example,
and
you
do
a
lot
of
computations
of
entropy
in
those
turn
out
to
be
correct,
so
there
can't
be
any
real
problem
with
it,
because
it's
consistent
with
empirical
observations,
but
despite
its
ubiquitous
appearance
in
textbooks
when
it
comes
to
foundational
questions,
there's
always
been
a
certain
air
of
mystery
around
the
second
law.
A
But
you
know
it
gets
into
some
things
about
the
history
of
heat,
which
sounds
a
little
strange,
but
like
it,
it's
actually
quite
fascinating
and
it
kind
of
goes
into
some
of
the
you
know,
early
concepts
of
heat
and
how
we've
really
kind
of
worked
from
the
concept
of
heat
to
the
concept
of
energy
and
then
heat
energy,
and
then
that
allowed
us
to
discover
the
second
law.
A
So
in
ancient
so
in
seeing
fire
one
might
imagine,
it
is
a
disembodied
form
of
heat
in
ancient
Greek
times
they
talked
about
everything
somehow
being
made
a
fire
and
also
somehow
being
intrinsically
in
motion.
So
these
are
metaphors
that
were
around
a
long
time
ago
before
they
knew
anything
about
molecules
or
the
second
law.
So
this
kind
of
is
built
upon
our
understanding
of
these.
These
processes-
and
it
kind
of,
goes
on
and
talks
about
the
more
recent
history
of
sort
of
the
things
leading
up
to
the
second
law.
A
So
last
week
we
talked
about
some.
What
was
the
topic
commensurability
in
the
philosophy
of
science
and
commensurability?
Is
this
thing
where
we
have
to
sort
of
have
you
know
the
cert,
the
ideas
in
our
science
sort
of
form,
formulate
how
we
come
up
with
theories,
so
our
Concepts
sort
of
lead
to
our
theories
and
that's
kind
of
what
he's
talking
about
here
where
some
of
these
concepts
of
like
what
heat
energy
is
and
how
it
works?
A
So
it's
really
interesting
kind
of
some
of
this
early
history
of
science
leading
up
to
the
second
law
and,
of
course,
it's
boltzmann
who
proposed
the
second
law
and
so
yeah
his
in
the
physicists
in
the
19th
century.
So
really
the
beginning
of
thermodynamics
itself
started
with
heat
engines,
and
so
the
Industrial
Revolution
started
off
and
one
of
the
sort
of
the
key
Innovations
there
was
the
steam
engine
where
efficient
steam
engines,
and
so
there
is
a
practical
concern.
A
So
you
know
it
goes
through
James,
Watt,
Carno-
and
you
know
some
of
these
things
that
they're
doing
with
machines
and
and
statistical
well
became
statistical
mechanics.
They
didn't
know
it
yet,
but
that's.
A
Going
and
then
there
was
this
also,
this
work
on
simultaneous
to
that
there
was
this
work
on
caloric
Theory,
which
was
of
course
not
which
was
not
correct
as
a
conceptual
framework,
but
this
idea
of
reversibility
and
operating
in
a
closed
cycle.
Those
things
were
correct,
and
so
they
ended
up
contributing
to
the
second
law.
Actually,
Colonel
introduced
the
second
law,
but
he
was
ignored
and
then
the
idea
came
up
again.
C
A
Was
being
devised
so
that's
yeah,
James
Clerk
Maxwell
actually
was
the
more
sort
of
famous
person
to
come
up
with
these
ideas
and
put
them
together
in
some
of
the
work
that
led
us
to
the
second
law,
so
that
yeah.
This
is
a
really
long.
Article
I
don't
want
to
go
through
the
whole
thing,
but
any
thoughts
on
that
they
didn't
talk.
Anything
about
biology,
though
really.
A
But
you
know
you
can
see
that
in
biology,
it's
been
applied
quite
extensively
in
terms
of
like
energy
consumption
and
in
terms
of
irreversibility
of
energy
and
things
like.
A
I,
don't
think
in
this
article:
let's
see
go
to
the
end
here.
It's
a
long,
article
I,
don't
think
in
this
article,
though.
Okay
now.
E
D
A
D
B
Systems,
you
have
to
be
careful
with
it
because
we're
basically
a
chaotic
system
and
if
you
tie
it
down
to
say
boltzman's
constant,
then
you're
not
not
looking
at
the
full
Behavior
or
something.
But
that's
that's
all
kind
of
off
to
the
side.
I.
B
What
you're
talking
about,
but
yeah,
it's
used
a
lot
in
in
seeing
if
the
system
is
stable
and
so
on,
I'm
I'm,
not
a
fan
of
necessarily
the
second
law.
Yes,
oh
well.
What
defies
entropy
and
I
say
life.
Does.
D
B
That
but
anyway,
yeah.
A
Yeah
so
well,
I
think
that's
the
part
that
we
didn't
talk
about
here
was
that
life
is
an
open
system
instead
of
a
closed
system,
and
so
it's
not
so
much
that
it
violates
the
second
law.
But
it's
definitely
like
you
know.
We
need
a
new.
We
need
to
think
about
it,
a
little
bit
differently
when
we
think
about
life.
A
Is
there
always
influxes
of
energy
coming
in
and
of
course,
you
get
structure
built
from
entropy,
but
you
also
get
structure
built
from
flows
of
energy
coming
in
so,
and
there
were
a
lot
of
things
like
dissipative
structures
and
other
types
of
Concepts.
That
came
a
little
bit
later.
A
A
So
this
is
peace,
peace,
I'm
really
trying
to
make
the
most
of
the
epigenetic
landscape,
and
so
this
is
a
picture
of
a
mountainous
landscape
or
rugged
landscape,
and
they
talk
about
development
here
and
they
talk
about
Conrad
Waddington
and
for
people
who
aren't
familiar
with
Conrad
waddington's
work
he's
a
developmental
biologist
who
developed
on
what
you
know
wrote
a
number
of
books
back
in
the
mid
20th
century
and
he
wrote
a
a
book
at
least
one
book
on
this
idea
of
the
epigenetic
landscape,
which
they
call
the
Waddington
landscape.
A
The
idea
is
that
it
looks
like
this
surface
that
you
know
is
unjulating
and
that
you
start
at
the
top
and
you
take.
You
have
a
ball.
It's
usually
a
ball
and
it
rolls
down
the
landscape
and
it
rolls
into
these
valleys
what
they
call
can
of
canals
or
canalization,
which
is
where
the
ball
rolls
through
the
area
of
least
resistance,
which
are
these
valleys
and
then
down
to
the
bottom,
and
these
valleys.
Actually,
branch
in
a
tree-like
way
so
that
you
end
up
with
multiple
possible
endpoints
in
your
tree.
A
A
So
the
problem
is,
is
yeah,
it's
highly
again,
High,
something
that's
highly
like
abstract.
The
ball
is
supposed
to
be
developmental
potential,
but
it's
kind
of
like
you
know
it's
not
that's.
That's
not
the
greatest
analogy.
So
so
it's
very
vague,
but
you
know
that's
a
lot
of
in
biology.
We
have
a
lot
of
conceptual
models
like
that,
which
is
maybe
good
and
bad,
because
yeah.
A
Oh
yeah,
great
okay,
yeah,
so
Waddington
was
the
last
Renaissance
biologist
and
he
he
had
a
lot.
He
had
a
background
in
philosophy,
paleontology,
embryology
and
genetics,
and
he
was
what
they
call
polymath,
which
means
he
had
a
lot
of
different
specialties
in
one
place,
and
so
he
one
of
waddington's
most
enduring
gifts
to
biology,
is
not
a
fact
hard.
One
throws
elegant
series
of
experiments.
A
Instead,
it
was
an
analogy
derived
from
an
understanding
of
the
data
available
to
him.
An
intuitive
notion
of
embryonic
development
should
proceed,
and
so
he
actually
talked
about
the
key
to
understanding
the
building
of
an
embryo
is
to
see
it
as
a
gradual
process
of
specialization.
A
There
is
life
each
so
undergoes
a
series
of
choices
with
time
the
choices
become
more
specific,
most
irreversible.
So
he
talks
about
this
irreversibility.
This
is
kind
of
like
an
entropy
when
you
get
this
irreversible
process
of
entropy,
and
you
know
the
need
for
more
energy
to
be
supplied
to
move
it
forward.
So
to
visualize
this
process.
He
came
up
with
this
hypothetical
landscape.
A
The
cells
of
the
early
embryo
can
be
thought
of
as
balls
as
at
a
high
point
in
the
craigie
OR
eroded
landscape
and
as
the
balls,
which
are
these
analogies
for
developmental
potential
roll
downhill,
they
will
enter
a
system
of
ranching
valleys.
There
are
many
possible
routes
down
the
hill.
Each
fork
in
the
valley
system
represents
a
decision
point
for
the
cell,
so
that
you
know
there
are
different
decision
points
in
development.
A
You
know
some
of
them
are
sort
of
the
differentiation
of
two
things,
so
endoderm
versus
ectoderm
foregut
versus
hindgut,
pancreas
versus
liver.
So,
as
differentiation
occurs
in
the
embryo
which
it
starts
out
as
this
undifferentiated
mass
of
cells,
you
get
these
points
where
things
get
differentiated
and
that's
those
are
these
division.
These
sort
of
branching
points
in
these
canals,
so
the
ball
can
go
down
one
way
or
another.
So.
D
A
Is
a
visually
appealing
metaphor,
but
then,
of
course
we
don't
know
I
mean
that's
just
a
metaphor.
Waddington
then
wrote
a
book,
the
strategy
of
the
genes
where
he
laid
out
sort
of
more
detail
about
the
underside
of
this
model.
Where
you
have
this
these,
they
almost
look
like
strings
attached
in
different
ways
to
the
bottom
of
this
model,
and
these
are
actually.
This
is
a
view
of
Gene
function.
So
you
have
these
genes
that
are
being
expressed.
A
Is
rolling
in
One,
Direction
or
another,
so
this
is
how
this
is
how
he
kind
of
you
know.
This
is
really
kind
of
pre-genome
pre-human
genome,
especially-
and
this
is
still
analogous
to
gene
expression
and
sort
of
a
view
of
Gene
function,
but
it's
actually
not
very
specific
in
terms
of
what
we
know
today
so,
but
it's
still
interesting
because
you
can
get
a
lot.
The
basic
structure
of
Gene
function
is
correct.
A
I
think
in
that
there
are
a
lot
of
interactions
just
the
way
you
know
putting
like
specific
things
on
these
boxes,
so
these
black
things
are
genes.
These
black
lines
are,
you
know,
different
ways
that
they
can
be
expressed,
and
then
they
underpin
these
canals.
So
it's
all
very
now.
You
know
it's
all
a
bunch
of
analogies,
but
that's
what
you
need
in
theory.
To
really
have
a
successful
theory.
Is
you
need
a
very
generalizable
model.
E
A
A
And
then
Susan
put
something
in
the
chat
here,
so
she
had
this
citation
here
from
2020
modeling
and
simulation
attend
security
structure
based
on
some
mechanics.
So
this
must
be
the
one
on
the
random
on
the
random
tensegrity
networks
or
10
security
structures.
B
No
that's
just
model.
A
A
And
then
the
reference
for
the
Triangular
tensegrity
is
in
my
PowerPoint
that
I
presented
and
sent
see.
A
B
I
showed
it
to
you,
that's
the
one
where
they
had
elastic
strings
and
then
stiffer
strings
and
the
stiffer
strings
produce
the
J
curve.
That's
more
like
a
cell
produces.
D
B
Anyway,
this
there's
a
form
Finding
method
for
tensegrity
structure
in
the
thesis
here.
The
modeling
simulation
with
the
semiomechanics,
of
course,
like
type
I.
B
Yeah
I
just
kind
of
interesting:
they
actually
built
a
physical
model
and
put
a
motor
on
it.
It's
just
it's
interesting.
You.
A
Well,
yeah.
That
brings
up
an
interesting
question
like
how
stable
are
they
in
motion
like
if
you
have
something
moving
through
a
medium
of
like
water,
or
you
know
air,
or
something
like
that.
You
know
it's
like
we're
testing
it.
This
I
guess
we're
testing
it
statically,
maybe
dynamically
we're
not
testing
it
like
in
a
environment
where
an
organism
might
move
yeah.
B
He'll
yeah,
he
can
give
me
Clues.
It
helps
actually
just
to
keep
them
in
mind
like
if
somebody
asks
well
what
about
the
sides
of
the
cell.
Aren't
they
contributing
to
things?
Well,
he
says:
there's
what
is
it
hydrolonic
acid
in
there?
That's
very
slippery.
That
has
no
friction
so
really
they're.
They
can
the
cells
themselves
themselves
control
their
attachments
and
they're
in,
like
they're,
the
acting
adherent
attachments
and
the
integrine
attachments.
A
If
anyone
else
said
anything
wanted
to
say
other
you
know,
I
can
say
now
otherwise,
okay
now
I'd
like
to
talk
about
something
that
we
published
in
2021
and
then
I
want
to
talk
about
like
a
follow-up
to
that
in
a
recent
paper
and
it's
this
concept
of
Developmental
tempo,
and
so
let's
go
to
the
first.
First
of
all,
this
is
the
paper
we
published
in
2021
and
it
was
published
in
biosystems
and
the
title
of
it
was
periodicity
in
the
embryo
emergence
of
order
in
space,
diffusion
of
order
and
time.
A
This
is
what
it
was
saying
and
Jesse
parent
as
well
as
myself.
These
are
all
collaborators
in
Devon,
so
the
abstract
reads:
those
embryonic
development
exhibit
characteristic
temporal
features.
A
This
is,
of
course,
a
parent
in
evolution,
and
this
was
something
that
Stephen
J
Gould
talked
about
with
respect
to
the
Temple
of
evolution
or
the
tempo
of
phylogeny,
and
so
this
means
that
change
occurs
in
bursts
of
activity
where
the
change
occurs
at
a
certain
sort
of
cadence
across
time,
and
so
what
we're
trying
to
do
here,
instead
of
looking
at
Tempo
and
evolution,
we're
looking
at
Tempo
and
development,
and
so
it's
a
very
similar
concept,
but
not
precisely
the
same
concept.
A
I,
don't
want
us
to
be
criticized
for
saying
that
you
know
we're
looking
at
evolutionary
tempo,
sensu
Gould
when
we're
actually
not
doing
that.
But
in
any
case,
what
we
do
in
this
paper
is
use
two
animal
models,
the
nematode
which
we
talked
about
in
the
group,
all
the
time
and
zebrafish
and
that's
another
model
organism
and
it's
a
very
different
organism
in
terms
of
its
development.
A
We
also
use
simulated
data,
which
is
a
numeric
simulation
of
sort
of
this
Tempo
idea,
taking
the
idea
and
applying
it
to
the
branching
structure
of
a
lineage
tree.
So
we
use
those
three
sources
of
data.
In
doing
so,
we
demonstrate
the
temporal
heterogeneity
exists
in
embryogenesis
of
the
cellular
level
and
may
have
functional
consequences,
so
proliferation
and
division
from
cell
tracking
data,
which
is
used
for
zebrafish
or
nematode,
is
subject
to
analysis
to
characterize
specific
features
in
each
model.
A
A
So
this
is
a
comparative
paper
that
uses
of
course,
data
from
C
elegans
data
from
Daniel
ririo,
which
is
the
zebrafish
and
a
numeric
embryo
which
we
have
created
so
the
C
elegans
data
and
then
zebrafish
data
are
both
things
that
we've
acquired
from
the
system,
science
and
biology
database,
and
so
this
is
based
on
Cell
tracking
data
for
C
elegans,
the
bowed
bio
waterl
paper
in
2006,
which
is
basically
the
precursor
to
the
Epic
data
set.
A
But
you
know
this
is
data
where
all
the
cells
are
tracked
and
development
in
C
elegans
and
then
the
Danny
ariri
or
zebrafish
data
set
is
based
on
the
cell
cell
tracks
of
nuclei
and
Keller
at
all
2008.
So
they
created
a
data
set,
that's
a
standard
for
zebrafish
cell
position
and
development.
A
A
A
We
also
used
numeric
embryo,
which
was
where
we
generated
pseudo
data
based
on
a
lineage
tree,
which
we
would
be
able
to
characterize
in
C,
elegans
and
zebrafish
by
cell
divisions.
We
do
a
similar
thing
here.
We
generate
pseudo
data
and
each
numeric
embryo
consists
of
one
or
more
vectors
describing
rounds
of
cell
division,
and
then
we
can
use
that
to
build
a
graph
to
show
the
compare
it
to
biological
data,
but
also
play
around
with
it
in
different
ways
to
determine
the
tempo
of
development
and
what
that
looks
like
in
terms
of
cell
divisions.
A
So
the
results
here
are
where
we
compare
C
elegans
with
zebrafish.
It's
obviously
a
very
different
type
of
different
tempo
of
development
and
C
elegans.
Its
development
is
completed
well
before
a
thousand
minutes
of
Developmental
time
and
zebrafish.
We
not
only
have
a
longer
developmental
period,
but
we
have
longer
rounds
or
longer
bouts
of
cell
division,
and
this
is
just
a
mark
because
it's
a
more
complex
phenotype
and
it
gives
you
this
pattern
where
you
get.
A
This
burst:
initial
burst
of
cell
division,
which
is
a
little
bit
off
center
from
what
you
see
in
C
elegans.
But
it's
very
you
know
it's
very
crowded
with
cells,
and
then
these
are
histograms.
So
they
show
the
frequency
of
cell
division
at
a
certain
point
in
time.
But
you
also
have
these
rounds
of
cell
division
later
in
development
between
1200
and
2000
minutes
in
between
twenty
four
hundred
and
three
thousand
minutes.
A
So
this
this
points
to
some
of
the
differences
in
the
embryo
and
the
biological
complexity,
the
organisms,
but
you
do
see
some
commonalities
as
well,
and
so
this
is
the
interval
between
cell
division
events
and
embryonic
development.
You
can
see
that
there's
a
larger
number
of
events
that
happen
rather
quickly,
so
you
have
cell
division,
events
that
are
rapid
and
then
you
have
less
rapid
cell
division
events.
There
are
a
couple
outliers
here
where
it
takes
almost
the
entire
duration
of
development.
A
C
A
As
to
the
you
know,
the
sort
of
distribution
of
of
speed
of
this
process-
and
you
see
the
same
thing
here
for
this-
is
I-
think
for
zebrafish.
So
this
is
where
you
get.
This
is
a
this
wind
red
window
is
this
is
a
periodic
region
in
a
periodic
region.
A
So
basically,
this
is
for
up
to
the
gastrulous
stage
in
development,
and
so
this
is
a
total
number
of
cells
and
it
goes
out
to
like
3
500
cells,
and
the
number
of
new
cells,
of
course
looks
like
they're,
these
distinct
rounds
of
cell
division,
up
till
about
1500
minutes
and
or
up
to
about
1500
cells,
I'm,
sorry
and
then
after
the
1500
cell
stage.
Then
you
get
these
more
aperiodic
division,
events
that
have
you
know
some
structure,
but
we
look
at
structure
we'll
look
at
like
kind
of
the
unitary
nature
of
these
bursts.
A
So
this
is
kind
of
a
you
know
without
using
like
high-end
machine
learning,
we
can
segment
this
time
series
in
this
way,
and
so
we
draw
a
box
roughly
around
this
periodic
set
of
stages
of.
A
This
is
just
a
statement
to
show
that
there's
this
transition
between
these
periodic
bursts
to
these
more
aperiodic,
bursts
of
cell
division
and
the
number
of
new
cells
that
occur
at
each
interval
of
development.
So
you
usually
have,
when
you
say
something
goes
from
an
A
to
a
16
cell
stage,
the
cells
double
all
the
cells
divide,
but
in
some
of
these
embryos
you're
going
to
have
asymmetric
division
events
and.
A
A
This
talks
about
numbers-
this
is
this
graph-
is
the
relative
frequency
of
division
rate
across
developmental
time
in
zebrafish.
This
is
in
zebrafish,
looking
at
the
frequency
of
division
rate.
This
is
analogous
to
this
other
graph
that
we
showed
for
for
C
elegans
this
one.
Here,
it's
just
a
different
way,
because
we
have
to
do.
A
Differently
for
zebrafish
for
a
number
of
reasons,
but
basically
we
have
a
similar
situation
where
we
have
cells
that
divide
quickly.
A
large
number
of
the
cells
divide
rather
rapidly,
so
you
have
the
cell
division
events
that
are
rapid,
that
maybe
go
over
two
or
three
minutes,
and
then
you
have
a
few
cells
that
actually
undergo
a
much
longer
period
in
between
their
division
times.
So
we
go
from
the
Mother
cell
to
the
daughter
cell
and
how
many
you
know
how
many
minutes
or
what's
the
sampling
interval.
A
And
then
we
look
at
what
we
we
do:
a
lot
of
work
on
embryo
networks
which
are
extracting
networks
out
of
embryo
tracking
data.
So
we
look
at
the
distance
between
the
cells.
We
look
at
the
position
of
the
cells
and
we
can
build
these
spatial
networks
that
are
called
embryo
Networks,
so
these
embryo
networks
would
characterize
them
for
zebrafish,
and
this
is
figure
six.
So
if
we
go
down
to
this
figure
five,
this
is
just
the
interval
size
of
Peaks
and
cell
division
for
all
developmental
cells.
A
This
is
again
another
transformation
of
the
original
data,
so
the
zebrafish
is
actually
notable
and
that's
very
different
from
c
elegans
c
elegans
forms
is
sort
of
mean
this
normal
distribution
with
a
mean
at
about
four.
It's.
A
More
off
in
terms
of
a
normal
distribution,
but
we'll
take
it
as
whereas
zebrafish
you
have
the
multimodal
distribution
in
terms
of
the
interval
size
in
terms
of
minutes,
so
there's
a
lot
more
diversity
in
terms
of
the
timing
of
zebrafish
embryos,
and
that
leads
us
to
this
sort
of
network
here,
where
we
have
this
embryo
Network
at
the
239
cell
stage
and
the
way
this
works
is.
This
is
not
like
in
any
sort
of
space
of
the
embryo.
A
This
is
showing
the
interactions
between
cells
just
to
show
that
if
they're,
you
know
more
connections,
you'll
see
more
lines.
So
it's
more
like
spaghetti.
If
there
are
more
connections
between
cells,
what
you
see
here
is
is
local
clustering,
so
that
makes
sense,
because
we
know
that
the
cells
are
segregating
and
we
think
that
these
bursts
or
these
differences
in
interval,
and
also
the
bursts
in
one
of
the
previous
graphs,
relate
to
this
sort
of
this
sort
of
structure
in
the
graph.
A
So
you
have
these
clusters,
these
local
clusters,
which
are
half
of
their
signatures
in
the
timing
of
division,
and
you
can
see
them
in
the
embryo
Network.
You
can
also
see
a
three-dimensional
representation,
so
this
is
an
embryo
Network
that
is
spatially
just
discrete
or
spatially
explicit,
and
you
see
that
these
blue
and
red
and
black
dots
all
have
let's
see.
So
this
is
sales
and
developmental
location,
color
coded
by
status
in
the
network.
A
The
white
cells
are
all
cells,
not
above
the
threshold,
so
we
threshold
our
connection
potential
connections
between
cells
and
they
have
to
be
in
within
a
certain
threshold.
The
distance
to
our
show
to
be
part,
you
know
to
be
actually
considered
to
be
connected,
so.
A
A
The
blue
cells
are
all
destinations
cells
with
one
Edge
to
another
cell,
so
the
red
cells
are
Source
cells.
The
blue
cells
are
destination
cells,
which
means
that
this
is
a
directed
Network.
A
The
red
cells
are
providing
a
connection,
or
you
know
in
terms
of
time,
the
blue
cells
are
destination,
cells
of
at
least
one
Edge
to
another
cell,
and
then
black
is
all
cells
of
more
than
eight
edges
to
their
cells,
so
black
ones
are
really
the
ones
to
pay
attention
to
these
are
the
hubs
in
this
network
and
red
and
blue
cells
are
basically
just
the
peripheral
nodes,
but
you
can
see
the
black
cells
are
clustered
in
a
certain
spatial
area.
These
are
the
Hub
cells.
A
So
then,
finally,
we
get
into
the
numeric
simulations.
This
is
where
we
have
the
number
of
Divisions
versus
developmental
time.
So
we
have
this
first,
this
blue
function,
which
is
I,
think
a
random.
This
is
a
uniform
distribution,
so
the
uniform
distribution
is
in
blue
and
it's
like
a
staircase.
It
just
goes
up
at
regular
intervals,
and
this
is
just
a
cell
lineage
tree
that
divides.
You
know
regular
intervals
and
that's
deterministic,
so
we
end
up
with
the
staircase.
A
By
contrast,
we
want
to
look
at
other
things.
We
don't
want
we're,
not
really
interested
in
the
normal
distribution.
The
standard
case
we
want
to
see
some
variation.
We
want
to
see,
lags
and
things
that
shoot
ahead
and
so
forth.
So
we
actually
have
modeled
this
with
other
types
of
distributions,
an
exponential
distribution
in
Orange,
which
shows
a
sort
of
staircase
the
poisson
which
is
gray.
That's
just
you
know
another
kind
of
version
of
exponential,
it's
not
really
exponential
but
and
then
binomial,
which
is
yellow
and
that's
another
type
of
distribution.
It.
A
Different
distributions
pull
away
from
this
normal
distribution,
and
so
we
can
actually,
you
know,
think
about
this
as
like
these
other
types
of
distributions
as
the
biological
case,
as
opposed
to
the
normally
distributed
case,
the
normally
distributed
case
would
be
sort
of
like
the
null
hypothesis
and
the
others
would
be
sort
of
a
header
behind
at
certain
points
in
development,
which
is
something
we
would
expect
to
see
from
a
biological
context
and
for
the
reasons
that
we
showed
there's
diversity
between
organisms,
they're
different
processes
that
are
spatially
explicit,
that
are
temporarily
explicit-
that
are
playing
a
role
here.
A
A
I
think
that's
it.
So
that's
that's
the
whole
study
that
we
kind
of
talk
about
these
simulated.
These
simulations.
We
talk
about
them
in
terms
of
organismal
outcomes,
so
we
didn't
plot
this
against
any
organismal
systems,
but
we
did
plot
this
sort
of
as
a
way
to
show
and
I
mean
if
you
look
at
those
graphs
in
the
previous,
the
results
in
the
previous
graphs
and
understand
that
these
sort
of
these
alternate
distributions
are
sort
of
that
better
fit
for
the
data,
and
so
that's
all
we
had
for
that.
A
That's
basically
the
idea
of
temp
Tempo
and
development
and
how
this
might
work
in
a
whole
organism
and
then
also
the
sort
of
the
computational
underpinnings
or
the
numeric
underpinnings
of
that.
So
this
is
a
paper
I
wanted
to
get
to.
This
is
a
more
recent
paper,
a
stem
cell
Zoo
under
uncovers
intracellular
signaling
of
Developmental
Tempo
across
mammals,
and
so
our
embryo
networks
actually
do
focus
a
bit
on
intracellular
signaling,
because
the
idea
is
that
you
have
these
areas.
A
That
you
have
the
you
know
the
distance
between
cells,
sort
of
governs
the
intracellular
signaling
that
occurs,
and
that
sets
developmental
Tempo,
but
it
also
sets
spatial
interactions,
and
so
there
are
a
lot
of
interesting
things
going
on
there.
So
the
abstract
reads:
differential
speeds
and
biochemical
reactions
have
been
proposed
to
be
responsible
for
the
differences
in
developmental
Tempo
between
mice
and
humans.
A
So
now
they're
looking
at
mice
and
humans
and
the
difference
there,
and
so
people
have
looked
at
developmental
Tempo
and
mammals
and
they've
looked
at
this
sort
of
idea
of
cell
division
and
biochemical
reactions
and
but
they're
not
taking
a
computational
view
here,
they're
taking
biochemical
in
a
mechanistic
view.
A
However,
the
underlying
mechanism
controlling
the
species-specific
kinetics
remains
to
be
determined
so
now
they're
going
to
be
looking
at
species-specific,
Kinetics,
and
so
this
is
going
to
be
a
study
of
pluripotent
stem
cells
and
they're,
going
to
be
looking
at
segmentation,
clocks
of
diverse
mammalian
species,
segmentation,
clocks
and
diverse
mammalian
species
during
your
body
weight
taxa,
so
that
includes
marmosets
rabbit
cattle
rhinoceros
together
at
the
mouse
and
human.
The
segmentation
clot
periods
of
the
six
species
did
not
scale
with
animal
body
weight
but
rather
grouped
according
to
phylogeny.
A
A
If
we
back
up
a
little
bit
and
think
about
like
what
scaling
is
generally
in
biology
or
in
this
type
of
biology,
is
that
you
have
scaling
that
occurs
according
to
body,
size,
body,
weight
and
then
metabolism
might
have
a
linear
relationship
with
body
size
or
body
weight.
So
we
expect
that
things
like
timing
for
development
might
be
tied
to
the
size
or
the
weight
of
the
organism.
So
for
small
organisms,
for
example,
they
have
a
higher
metabolism
than
larger
organisms,
and
so
we
expect
this
for
the
segmentation
block.
A
But
we
don't
actually
see
this
with
animal
body
weight.
We
see
this
with
respect
to
phylogeny,
suggesting
that
there's
a
common
ancestry
for
certain
tempos,
the
biochemical
kinetics
of
the
core
clock,
Gene
hes7,
which
is
a
clock
mechanism,
so
cell
division
is
regulated
by
sort
of
clock
mechanism
or
this
time
keeping
mechanism.
It's
a
circuit
that
regulates
the
clock
of
cell
division.
So
there
are
a
number
of
stages
of
cell
division
that
have
to
happen
in
sequence
and
sometimes
they
get
slowed
down.
A
So
this
Gene
is
actually
responsible
for
a
lot
of
this.
However,
the
cellular
metabolic
rates
did
not
show
an
Evidence
correlation.
Instead,
genes
involving
biochemical
reactions
showed
an
expression
pattern
that
skills
of
the
segmentation
clock
period
so.
D
A
Means
that
expression
patterns
certain
genes,
skills
of
the
segmentation
clock
period,
meaning
that
there's
this
relationship
between
gene
expression
in
this
period
and
then
the
developmental
Tempo,
all
together,
are
stem
cell,
uncover
General
scaling
laws
governing
species-specific
developmental
Temple.
C
A
And
it's
usually
like
one
set
one
one
variable
over.
D
A
So
this
could
be
like
body
length.
A
And
this
could
be
like
some
fact
or
I
guess
why?
Because
that's
actually
the
y-axis,
and
so
this
would
scale
so
for
different
body
lengths.
You
get
different
values.
A
Usually,
linear,
it's
usually
log
linear,
you
know.
Usually
we
get
like
a
logarithmic
function
where
things
kind
of
tail
off
with
the
upper
and
the
body
length,
but
it's
usually
controlled
by
one
variable
over,
like
you
know
so
like
for
different
values
of
body
length,
you
get
different
values
of
this
x
variable,
and
so
you
know
for
the
clock
you
might
have
like
you
know,
for
metabolism,
for
example,
y
might
be
metabolic
rate,
and
this
might
be
body
length
and
you
can
basically
say
for
different
body
links.
You
have
a
different
metabolic
rate.
A
To
Define
these
variables,
that's
why
I
use
y
here,
but
for
Tempo
you
might
have
like
a
longer
Tempo
or
higher
rate
of
change
or
lower
rate
of
change,
depending
on
how
you
set
this
graph
up
with
respect
to
body
length
and
what
they
find
action
is
paper.
Is
that
it's
actually
a
matter
of
phylogenine,
not
body
length,
which
means
that
phylogeny,
of
course,
is
very
different.
A
A
And
if
you
think
about
this
in
terms
of
like
a
cell
division
clock,
you
would
acquire
certain
genes
and
they'd
be
in
a
circuit
and
they'd
have
the
certain
tempo
of
cell
division.
So
we
actually
want
to
look
at
this
one
and
then
that's
conserved
across
the
descendants
of
the
common
ancestor
with
some
variation.
A
So
these
you
know
these
genes
in
the
circuit
might
have
exhibit
mutations
and
different
descendant
organisms
in
the
tree
and
they
might
have
different
rates
of
change.
So
the
rates
of
change
might
be
variable
here,
but
the
circuit
will
remain
intact
for
the
most
part
and
for
example,
but
but
also
in
this
Quade,
you
might
have
a
different
set
of
genes
that
result
in
a
different
Tempo
and
again,
you
may
have
this
local
variation
within
the
circuit,
but.
C
A
So
this
paper
kind
of
focuses
in
on
stem
cell
and
pluripotent
cell
Technologies
they're
using
stem
cells
from
diverse
Rebellion
species,
they're,
actually
looking
at
embryonic
stem
cells
and
induced
potentials,
but.
D
A
Is
of
course,
these
are
cells
that
are
Immortal,
meaning
they
can
divide
infinitely.
These
are
cells
that
are,
you
know
they
have
this.
A
You
know
we
think
about
stem
cells
as
being
sort
of
you
know
what
they
call
pluripotent,
which
means
that
they
have
the
potential
to
take
on
many
Fates.
A
But
of
course
there
are
a
lot
of
different
types
of
stem
cells,
so
that
pluripotency
is
limited
to
maybe
like
any
kind
of
embryonic
cell
or
any
you
know,
neural
stem
cells
or
any
kind
of
neural
cell,
so
it's
not
exactly
any
kind
of
cell
and
so
in
in
the
case
of
stem
cells,
cell
divisions
are
rather
important,
as
we
saw
in
the
developmental
case
in
zebrafish
and
C
elegans.
A
You
have
this
you,
those
are
all
developmental
cells
or
different
forms
of
pluripotent
cell,
so
these
aren't
cells
that
have
differentiated
they're
sort
of
on
their
way
to
differentiation.
So.
D
A
A
good
way
to
look
at
cell
division
and
sort
of
this
Tempo
and,
of
course,
the
temporal
development
leads
to
all
sorts
of
different
structures.
All
sorts
of
different,
different,
diverse
phenotypes,
one
of
the
benefits
of
having
being
able
to
play
around
with
this
Tempo
in
development
is
to
end
up
with
different
types
of
specialized
structures.
A
For
example,
if
you
have
to
grow
a
brain
much
larger
than
you
would
normally
have
it
in
that
lineage.
You
know
there
might
be
some
modification
to
the
tempo
of
development
rather
than
the
actual
cells
themselves,
and
that
will
just
lead
you
to
to
having
more
division
around
the
division
and
a
larger
structure
overall
in
a
certain
part
of
the
body,
and
so
that's
why
we're
interested
in
this
this
paper
is,
you
know
it
kind
of
goes
through
this
idea
in
stem
cells
and
different
species.
A
F
A
Right
after
the
end
of
this,
so
all
right,
Ariel
thanks
for
attending.
If
you
have
any
questions
about
anything
gsoc
or
anything
related,
let
me
know
and
I'll
be
around
in
slack
and
email
and
all
that
so
have
a
good
week.
Thank
you.