►
From YouTube: DevoWorm Meeting #33: Diatom movement tracking, DDEs, inheritance and morphogenesis in C. elegans
Description
Discussion of mathematical modeling for Diatom movement tracking. Master plan for incorporating 2022 GSoC contributions into DevoLearn and DevoWormAI. A short tutorial on Delay Differential Equations (DDE). Papers on C. elegans transgenerational inheritance and gonad morphogenesis. Attendees: Alon Samuel, Bradly Alicea, Susan Crawford-Young, Morgan Hough, Anant Kumar, and Karan Lohaan
A
All
right,
I,
don't
know
who
else
is
coming
today,
but
I
know
dick
can't
make
it
because
he
has
a
problem
with
his
internet.
Oh,
and
there
are
other
people
who
were
doing
gsoc
and
G-Shock
is
over
so
I'll
be
taking
a
break.
Maybe
I
will
see.
Oh
yeah
yeah.
B
A
B
Yeah
because
I
thought,
like
maybe
there
was
a
convention
that
you
went
to
last
week
or
maybe
it's
gonna
happen,
I,
don't
remember
which
one
you
said
last
time.
A
Oh,
the
yes,
the
incf
UMC
assembly,
yeah,
well,
yeah
I,
went
to
that
and
that
was
actually
pretty
good.
They
had
a
number
of
different
top.
They
covered
a
number
of
different
topics.
I
talked
about
this
in
our
Saturday
meeting.
Morgan
was
there
where
I
talked
about
a
lot
of
the
neuroinformatics
stuff
that
they've
been
doing
so
they
do
a
lot
of
stuff
with
like
dealing
with
processing
data
once
it's
collected
and
making
that
available
to
people
making
it
public.
It's
it's
a
lot
of
work.
A
It's
not
just
something
you
can
just
throw
up
on
the
internet.
You
know
a
lot
of
naming
conventions
and
file
type
conventions
and
things
like
that
and
then
yeah.
You
know,
there's
some
other
work
on
yeah
there's
some
other
work
on
different
platforms
like
Neuroscience
neurodata,
Without
Borders.
A
So
this
is
something
that
they
use
in
open
worm
a
lot
and
it's
a
sort
of
a
convention
for
taking
larger
electrophysiology
data
and
processing
it
into
something
that
can
be
shared
and
worked
with
in
you
know
anyone
can
work
with
it.
If
you
have
the,
if
you
have
the
right
things
formatted
correctly
and
then
you
know,
you
know
what
you're
doing
so,
that's
all
you
need
and.
A
We
do
a
lot
of
that
in
this
group
with
a
lot
of
secondary
data
that
we
use.
We
don't
have
as
high
a
level
of
conv
of
sort
of
formality
is
what
they're
doing
but
I
like
to
cut
it
we're
a
little
bit
simpler,
but
that's
that's.
That
was
what
that
was.
B
So
they
they
use.
You
said
that
they
use
Diva
worm
as
what
is
a
simulation
basis
and
they
also
like
oh.
A
No,
what
they
do
they
have
this
thing,
Neuroscience
Without
Borders,
it's
a
way
to
handle
data,
so
open
worm
does
a
lot
of
stuff
with
data.
It's
it's
a
lot,
there's
a
heavy
emphasis
on
simulation,
but
the
inputs
are
always
some
sort
of
neural
data,
and
so
they
have
to
use
different
sort
of
platform.
You
know
different
types
of
software
to
bring
data
into
the
simulation,
so
you
can't
just
simulate
nothing.
You
have
to
have
data
to
simulate
with
yeah.
So
that's,
okay,
that's
where
Neuroscience
neurodata
without
quarters
comes
in.
A
Yeah,
we
don't
really
use
it,
because
we,
a
lot
of
our
secondary
data,
is
developmental
biology.
So.
A
B
Yeah,
don't
I
don't
have
much
to
update
I've
been
going
through
the
equations
and
some
of
the
background
that
we
talked
with
Thomas
about
the
tracking
and
the
method
to
to
Define.
If,
if
the
movement
is
jerky
or
smooth
and
I've
been
having
some
questions
about
it,
and
if
we
gonna
have
time,
maybe
it's
kind
of
like
a
it's
gonna,
be
nice
to
write
some
equations
down.
I,
don't
know
if
they
also
like
the
technology.
B
Gonna
allow
it
to
write
it
in
I,
don't
know
like
a
board
or
something
yeah
I'm
sure
it's
gonna
allow
it,
but
just
like
if
it's
gonna
be
like.
You
know
like
fast
enough
like
for
me
to
write
it.
B
What
I
thought
about
and
then
like
to
maybe
discuss
it
with
you
yeah,
because
there's
something
that
I'm
I'm
kind
of
like
I
wasn't
sure
that
I'm
agreeing
with
the
kind
of
some
of
the
conclusions
so
I
was
I
was
just
like
wondering
like
to
make
it
more
clear
for
myself,
because
that's
one
and
another
one
was
to
I
just
started
to
develop
some
kind
of
maybe
like
tools
for
myself
to
just
to
do
the
diatom
tracking.
B
Let
me
show
just
a
picture,
for
example,
share
it.
You
have
three
options:
maybe
it's
all
right,
so
maybe
I'll
try
and
I
can
talk
about
it.
So
there's
like
a
passillary,
it
kind
of
consists
of
like
different,
like
diatoms,
it's
kind
of
like
moving
like
in
the
together
yeah
and
the
problem.
B
The
video
to
because
some
of
the
diatoms
are
kind
of
like
going
out
of
the
of
the
frame
and
back
kind
of
like
down
so
just
to
crop
them
to
get
to
to
have
like
a
datum
that
kind
of
moves
within
the
frame
and
not
like
gonna
I'm,
not
gonna,
lose
it
going
too
much
and
out
of
the
frame
yeah.
So
just
just
doing
that
at
the
moment.
So
just
initial
work.
A
Yeah,
that's
that's
good.
Yeah
I
wish
you
could
have
seen
what
you
were
doing,
but
that's
the
way
it
goes
sometimes.
B
B
Yes,
so
let
me
can
we,
is
there
a
board
or
something
too.
A
I
have
a
board,
but
it's
I
have
it
on
my
end,
but
I
I,
don't
know
if
you
want
I
mean
I
can't
share
it.
So
let
me
share
myself
or
you
could
yeah
I.
Don't
know
we
tried
this
before
with
the
sugar
screen.
B
Yes,
that's
that's,
but
if
I'm,
if
what
I'm
gonna
do,
if
I'm
gonna
do
like
an
online
board
and
then
you
can
maybe
share
it
and
I
can
show
you
the
link.
B
To
sign
up
all
right
foreign.
B
B
Okay,
so
we
yeah
it's
gonna,
do
a
pen.
Yes,
so
we
thought,
let
me
bring
my
equations
okay,
so
we
said
that
x
like,
for
example,
X
that
we
tracking
X
each
X
is
the
point
in
kind
of
like
a
time
yeah.
B
So
that's
gonna
be
X
of
T
of
time.
That's
going
to
be
roughly
from
two
components:
that's
going
to
be
x,
s,
1,
plus
X,
okay,.
B
Of
one
and
then.
B
B
And
sorry
about
the
writing,
yeah
so
and
X2
I
was
kind
of
thinking
he
of
tea
that
it's
X
as
to
plus
X
J2,
but
that's
also
yeah.
So
that's
x,
one
plus
some
sort
of
like
this
kind
of
like
L,
that
can
kind
of,
like
maybe
yeah,
measure,
right
length,
SL.
B
Yes,
the
length
between
them
I
thought
that
maybe
because
we
don't
know,
there's
gonna
be
a
problem
with
measuring
it,
because
we
can.
We
can
maybe
have
a
like
an
accuracy
problem.
So
maybe
it's
going
to
be
with
some
sort
of
like
a
plus
or
con
which
you're
both
going
to
be
constant.
B
But
like
we're
not
gonna
know
if
we're
going
to
subtract
this
L
that
we're
going
to
measure,
maybe
there's
gonna,
be
an
inconsistent
because
it's
going
to
be
with
another
like
small,
inaccuracy
or
an
error.
So
maybe
this
hour
is
gonna
like
just.
B
Let
me
saying
about
this
small
error
really
so
be
more
on
Epsilon
yeah,
yes,
and
then
what
I
thought
that
so
there's
gonna
be
and
if
we're
going
to
track
both
of
them
so
from
X,
it's
going
to
be
like
a
t
tier
one,
so
we're
gonna
track
it
and
then
for
T1.
B
It's
gonna
be
like
this
X
1,
plus
a
sort
of
like
noise
that
it
calls
it
so
there's
gonna
be
a
noise
and
I
call
it
like
a
noise
that
is
mutual
to
both
of
the
kind
of
like
The,
Trackers
or
yeah.
It's
gonna
happen
like
for
each
tracking
that
I'm
gonna,
add
and
another
noise.
That's
unique
to
the
track.
B
That's
I'm,
gonna,
I
kind
of
just
gonna
call
it
n
one
and
I
I
wasn't
I,
wasn't
sure
like
because
he
was
more
saying
and
then
and
then
there's
gonna
be
a
T2,
that's
gonna,
be
like
X2
and
plus
and
and
this
Mutual
noise,
sorry
and
plus
noise
that
is
unique
to
this
tracker
and
and
I
was
I
was
thinking
that,
if
we're
gonna
correlate
between
t
one
and
two
so
there's
still
going
to
be
a
component,
that
is,
that
is
highly
correlational
because
of
the
the
the
noise
that
is
mutual
for
both
of
them
within
the
tracking
right
or
also
like
within
the
movement
of
like
X1
and
X2
they're,
also
like
within
it.
B
A
Well,
okay,
yeah,
so
the
mutual
noise
is
going
to
be
like
what
X1
and
X2
are
going
to
have
in
common.
So
if
they're
on
the
same
cell,
they
will
probably
be
other
well,
I
would
say:
they'd
probably
be
highly
correlated
just
simply
because
the
cell,
you
know,
there's
no
internal.
Well,
there's
a
little
bit
of
internal
movement
in
the
cell,
but
it's
largely
just
kind
of
like.
If
there's
a
Jitter
in
the
cell
say
if
it's
moving
up
and
down
kind
of
vibrating
at
the
end
of
its
track,
you
know
how
it
oscillates.
A
So
if
it's
at
the
end
of
an
oscillation,
it's
going
to
show
up
in
both
points,
so
you're
measuring
just
two
points
on
the
same
cell,
but
actually,
if
you
think
about
it,
when,
when
you
kind
of
move
from
like
the
end
of
the
cycle
to
the
to
moving
back
in
the
other
direction,
you
have
these
countervailing.
You
know
momentum,
you
have
these
countervailing
forces
that
are
inertial
forces
that
are
moving
it.
A
A
Some
correlation,
but
it
for
practical
purposes
at
any
one
time
that
is
going
to
be
decorated.
Now,
if
you
have
like
multiple,
let
me
see
if
I
can
do
this
if
you
have
multiple
cells.
So
if
you
have-
and
of
course
this
is
the
Practical
case
where
you
have
a
stack
of
these
yeah,
it's
kind
of
moving
and
you're
just
moving
in
this
direction.
You'll
have
these
two
points
on
each
cell
and
then
the
question
is,
if
you're
tracking,
between
cells,
because
that's
important
as
well.
You
know
this
point,
which
would
be
X1.
A
One
I,
don't
know
x.
One
two
excuse
that
notation
these
two
points
here
and
here
will
be
they'll,
be
correlated
as
well,
but
you'll
see
that
same
effect
where
if
this
one
hits
the
top
and
then
it
starts
to
move
in
the
other
direction,
this
one
will
follow,
but
it'll
be
at
a
lag.
So
there
will
be
a
lag.
A
So
I
think
that's
important
that
there's
like
even
with
noise
that
there's
some
sort
of
lag
involved,
because
it's
not
all
uniform
I
mean
it's
a
it's
sort
of
a
uniform
process
in
the
sense
that
at
the
macro
scale.
But
when
you
think
about
like
the
individual
measurement
points,
there's
this
effect
of
inertial
forces,
especially
at
the
ends.
And
that's
where
you
see
a
lot
of
the
jerkiness
as
I
recall,
is
at
the
ends
of
the
oscillations
where
it's.
B
Okay,
yeah
within
the
change
of
Direction,
when
the
movement
is
becoming
like
less
kind
of
like
maybe
like
Heim
in
Maybe
One
Direction,
just
when
it
changes,
yeah,
okay,
yeah,
so
I've
got
like
yeah
I
got
like
I,
didn't.
Think
we're
like
within,
like
different
diagrams
that
yeah
it's
gonna
be
a
lag
I've
got
like
maybe
reaching
yeah.
Some
point,
that's
kind
of
like
that's
good
and
I'm.
B
A
A
The
the
change
in
direction
is
pretty
Swift,
and
you
know
if
you're
at
the
end
of
like,
if
you
think,
if,
if
you
think
about
it
as
like
a
large.
A
Maybe
like
a
a
cruise
ship
or
something
where
you
have
someone
on
one
end
versus
the
other
and
you
change
direction.
You
know
they're
going
to
experience
it's
a
rigid
structure
but
you're
going
to
experience
different
inertial
forces
as
you're
moving
as
you.
If
you
change
your
direction
really
quickly,
you
know.
Is
this
going
to
be
a
different?
Yes
experience,
so
that
that
that's
okay,
yeah.
B
That's
good
yeah
and
yeah
I
was
I.
Think
later
on,
I
was
kind
of
like
going
to
the
maybe
some
of
the
Assumption
of
kind
of,
like
the
we
later
on,
talked
about
like
a
thing
within
the
presentation
about
the
expectation
the
mean
of
like
T1
and
then
like
the
main
of
T2.
B
But
I
didn't
go
as
much
as
that
afterwards,
but
I
think
I
can
now
it
kind
of
makes
more
kind
of
maybe
sense
to
kind
of
think
about
it,
as
maybe
like
a
delay,
and
then
I
can
maybe
continue
with
the
my
analysis
of
the
correlation,
which
is
I,
think
yeah
I
think
it's
something
like
the
correlation
of
t.
B
That's
perfect
and
then
that's
also
like
divided
I,
think
that
was
the
variance
of
T1.
B
So
I'm
gonna
continue
to
kind
of
like
analyze
it
but
didn't
get
as
far
as
as
there
right
so
I
think
now
it's
gonna
help
her
helping
me
more
I
am
going
to
be
interested
to
see
like
maybe
if
Thomas
Made
I
don't
know,
maybe
a
script
that
I
can
then
re-uh.
B
We
also
like
write
it
like
a
version
for
myself
like
on
the
python
code
and
then
like
produce
it
and
upload
it
to
the
to
the
repo
I
think.
That's
gonna
be
good
for
reproducibility
of
the
code
yeah,
and
that
would
be
good.
Okay.
Yes,
all
right,
yeah,
that's
I,
think
that
was
kind
of
maybe
what
I
wanted
to
discuss.
Yeah
presentation.
A
Yeah,
that
is
a
pretty
nice
demonstration
of
the
math
and
everything
they
have
hello,
Quran
yeah,
so
we
yeah
another
option
and
I
haven't
talked
about
this
in
the
group
very
much,
but
I've
done
some
work
with
it
in
the
past.
Our
delay
differential
equations.
Are
you
familiar
with
those.
A
It's
just
like
a
standard
differential
equation
and
they're,
usually
discrete
and
or
they
could
be
continuous
and
it
basically
just
introduces
a
delay
term.
So
the
idea
is
that
you're
measuring
something
in
time,
you're
measuring
the
function
and
you're
integrating
over
that
function.
To
see
you
know
what
the
change
over
time
is,
you
know
and
treat
it
as
a
dynamical
system,
but
you
also
have
this
the
wave
Factor.
So
there
are
different
ways.
A
You
can
model
things
with
the
way
and
it's
like
you
know
where
you
expect
something
to
be
delayed
for
a
little
bit
or
some
process
to
have
a
delay
in
it
and
that's
that's
usually
the
way
they
work.
It's
kind
of
a
neat
tool,
I,
don't
think
that
a
lot
of
yeah
I,
don't
think
a
lot
of
people
have
direct
experience
with
it,
though,
but
that's
just
this
problem
just
made.
A
B
Yeah,
it's
kind
of
like
it's
it's
exactly
like
there
because
of
this
delay
effect
of
this
well,
basically,
with
the
kind
of
like
this
motion
of
oscillation,
plus
kind
of
like
a
a
kind
of
like
effect,
that
kind
of
comes
from
the
physics
of
God,
like
the
diatom's
kind
of
like
I,
don't
know,
maybe
frictioning
between
each
other
or
like
something
that
happens
there.
We
are
done
yeah
yeah
exactly
now
between
between
them,
okay,
cool
yeah.
Thank
you.
A
So
if
you
want
to
look
for
software
packages
to
do
this,
don't
do
it
by
hand
dd45,
which
is
a
Matlab
package,
foreign
and
then
there's
also
od45
and
I.
Don't
know
if
od45
will
work,
I
mean
there's
a
special
parameters
being
passed
for
DDS
and
it
may
not
work.
Octave
has
od45
and
Matlab
as
well.
They
don't
know
if
octave
has
a
version
of
dd45,
it
might
be
I,
don't
know,
and
then
there's
Pi
delay,
which
is
for
python,
which
is
a
okay
python.
A
That
is
for
it's
a
you
know,
toolbox
that
you
have
to
install
and
I'm,
not
really
sure
the
quality
of
those
Solutions
but
they're.
Usually
those
are
the
three
main
ones
that
I
found
for.
So
we
have
the
one
for
Matlab
here,
which
is
this.
This
is
actually
for
sort
of
the
help
for
dd45,
there's
Pi
delay,
which
is
a
version
0.1.1,
there's
some
documentation
for
that,
and
so
that'll
help
you
through
some
of
this.
A
So
this
is
an
example
of
a
delay
differential
equation.
You
can
see
that
there's
a
some
sort
of
dynamical
system
here
where
you
have
a
Time
series
and
you
have
a
step
function
here
where
something
goes
as
glass
and
there's
a
step
function
then
there's
additional
Dynamics
and
then
it's
just
kind
of
maybe
noise
at
the
end.
So
you've
got.
This
is
actually
heart
rate
for
Barrel
Flex
feedback
mechanism.
This
is
another
problem
here,
the
D1
problem
for
of
and
right
in
Hayashi,
and
they
show
this
over
time.
A
So
basically,
the
way
differential
equation
differs
from
an
ordinary
differential
equation
in
a
number
of
ways,
so
like
an
ordinary
differential
equation
dealing
with
a
dynamical
system,
so
it
could
be
a
discrete
system
in
time
or
it
could
be
a
continuous
system
in
time
it
and
that's
going
to
depend
on.
You
know
how
you
treat
the
solution.
A
Okay
and
there
you
go
the
way
differential
equation
is
a
little
bit
different.
The
way
differential
equation
is
we
have
the
same
form.
It's
a
differential
equation
and
you're
solving
for
some
function
that
that
function
then
has
another
component,
which
is
the
way,
and
so
you
could
actually
have
like
this
sort
of
delay
here,
usually
use
a
Tau,
which
is
the
symbol.
A
A
That
sort
of
thing
so
you're,
basically
you
have
this
extra
term,
which
is
Tau
and
it's
usually
a
minus,
so
I
have
to
put
a
minus
sign
in
there,
so
you're
solving
for
this
function,
but
minus
this
delay,
and
so
what
does
this
look
like
in
practice?
Let's
say:
I
have
a
dynamical
system
here
and
I
have
a
curve.
A
It's
kind
of
going
like
this
and
I
want
to
know
like
say
that
I
assume
that
there's
some
sort
of
delay
in
this
curve
so
say
this
is
like
the
expression
of
some
gene
expression
factor
and
it
starts
to
go
up
with
time.
This
is
a
new.
This
could
be
solved
easily
with
an
ode
you
could
find
out
about
different
moments
of
its
time,
Evolution.
A
But
then,
let's
assume
that
we
okay
say
we
want
to
delay
it
somehow.
We
assume
that
there's
some
sort
of
delay,
maybe
there's
some
sort
of
drug
that
we
give.
We
treat
the
cells
with
or
something
and
there's
a
delay.
So
what
does
that
look
like?
A
So
that
means
you
have
this
flat
line
here
and
then
you
start
getting
Dynamics,
and
this
is
your
delay
or
your
Tau,
and
so
now
you're
going
to
take
this
OD
minus
the
Tau
to
figure
out
what
this
delay
looks
like
okay,
you
know
you
could
have
another
case
where
you're
trying
to
take
like
a
certain
interval
of
the
delay
so
like,
for
example,
this
interval
here.
A
Where
you
have
Tau
over
a
certain
range
and
you
can
subtract
it
from
the
OD
over
a
certain
range
and
you
can
get
your
solution
now.
There
are
different
types
of
things
you
can
use
this,
for
you
can
use
this
for
gene
expression.
You
can
use
this,
for
we
were
talking
in
the
meeting
about
movement
of
microorganisms
with
diatoms
looking
at
their
sort
of
higher
order,
Dynamics,
and
this
would
be
useful
for
that.
A
I
did
a
paper
along
about
eight
years
ago.
Now.
This
is
where
we're
looking
at
cells.
These
are
mammalian
cells
and
looking
at
the
degradation
of
RNA
based
on
certain
drug
treatments-
and
you
can
do
you
know
you
can
do
drug
treatments.
Very
treatments
are
a
good
way
to
shut
down
different
processes
in
a
Cell.
A
But
of
course
what
we
found
too
is
in
some
of
these
cell
lines
or
in
some
of
these
genes
that
we
looked
at.
They
were
sort
of
sequestering
RNA.
You
know
when
you
first
when
it
was
first
exposed
to
the
stimulus
it
was
sequestering
RNA
and
then
there
was
later
Decay.
So
this
is
a
nice
dynamical
system
that
you
can
model
and
use
delay
differential
equations
to
model,
so
in
figure
nine
of
this
paper.
A
Just
here,
let
me
zoom
in
you
can
see
how
we
kind
of
approach
this
so
This
Is
A,
first
order
DDE.
This
is
the
state,
and
then
you
have
all
these
different
types
of
responses
that
you
want
to
model.
So
the
delay
is
dependent
on
the
type
of
response
that
we're
observing
so
there's
aggregation,
there's
non-linear
Decay
and
there's
a
regular
Decay,
so
for
regular
Decay
the
way
is
zero
for
non-linear
Decay.
The
Decay
is
some
rate
and
then
aggregation.
A
Now,
let's
talk
about
the
different
types
of
system
that
you
might
have
or
the
different
types
of
the
way
you
might
have
so
the
first
type
of
delay
is
constant
delay,
and
so
what
I've
talked
about
here
in
the
let
me
State
this
all
right
so
and
and
in
that
last
example,
I
think
you
saw
an
example
constant
delay,
and
you
saw
an
example
of
how
it's
not
constant,
constant
delay
is
where
you
have
something.
A
That's
always
delayed
relative
to
your
ode,
so
you
know
you
might
have
like
something
that
starts
later
and
has
the
same
sort
of
growth
characteristics
or
it
might
be
some
break.
That's
put
on
the
process
at
all
times,
it's
constant,
so
you
don't
need
to
worry
about
doing
much
to
it
other
than
just
having
a
generic
delay
term.
A
A
So
time
dependent
delay
is
simply
what
time?
What
point
in
the
process
are
you
and
what
is
that
the
way
or
that
the
way
profile
look
like
so?
For
the
last
example
you
had
where
you
had
sequestration
of
RNA,
you
had
a
negative
delay.
You
know
there
was
super
linear
growth
and
then
later
you
had,
you
know
positive
the
way
so
e
or
Decay,
which
would
also
be
delay
in
that
group
in
the
growth
rate.
So
you
have
two
different
components:
at
least
non-linear.
A
You
know
non-linear
types
of
control
like
saturation
and
other
things.
Those
are
also
time
dependent
in
that
different
points
in
time
in
that
function,
there's
a
different
amount
of
delay
so
like.
If
we
hit
something
like
this,
we
had
a
peak
and
then
a
decline.
A
The
third
one,
of
course,
then,
is
State
dependence
and
that's
a
little
bit
different,
and
this
is
where
you
get
into
your
discrete
models,
so
you
don't
have
to
have
like
discrete
now
it's
a
discrete
dynamical
system,
which
is
a
DDE,
and
these
are
different
even
yet,
because
Odes
are
just
basically
ordinary
differential
equations
are
usually
continuous.
A
A
This
is
like
a
DD.
You
know
made
a
mistake
here:
okay,
so
this
is
a
DD
which
is
or,
and
then
this
is
a
actually
I-
think
that
they
get
the
way
they
do.
The
nomenclature
is
it's
still
an
OD,
but
it's
a
discrete
OD.
So
it's
just
basically
like
you
know.
You
have
different
states
and,
depending
on
the
state
that
it's
in
what
you
have
to
have
is
like
discrete
States
I
can
see
they
have
continuous
States.
A
A
And
so
that's
actually
something
like
it's.
It's
not
time
dependent
it's
state
dependent
because
you
could
be
in
a
state
at
any
interval
of
time.
You
want
it's
just
that
when
you
transition
on
that
state,
that
delay
is
what
you're
modeling.
So
your
you
know
your
towel
would
look
something
like
a
step
function
almost
over
time.
A
This
is
your
different
toes
and
then
that's
going
to
be
a
factor
in
your
in
your
solution,
and
so
it's
not
time
dependent,
but
it's
state
dependent
just
because
you're
in
a
different
state
and
you're
in
a
state
over
a
certain
period
of
time.
So
that's
a
right
confusing.
Then
you
have
something
else.
You
have
four,
which
is
a
little
bit
less
clear.
Maybe
and
those
are
jumps
in
special
events,
and
this
is
where
you
basically
have
these
kind
of
states,
but
they're
not
evenly
they're,
not
known,
maybe
or
they're,
not
evenly
distributed.
A
So
you
know
that
could
be
like
there
are
all
sorts
of
jump,
processes
and
Mathematics,
but
you
know
in
in
biology
as
well
jumping
from
one
state
to
another,
maybe
jumping
from
a
stem
cell
to
a
neuron.
There
is
no
difference
in
the
way,
but
if
there
were
that
would
be
a
jump
and
it
would
be
discontinuous
special
events
being
something
like
you
know,
if
you
think
more
broadly
than
biology
like
at
earthquake
or
some
type
of
like
you
know,
special
event,
like
you
know,
there's
some
catastrophe.
A
You
know
you
have
some
brain
injury
or
something
that's
that's
a
special
event
right.
So
you
can
measure
those
as
well-
and
it's
very
you
know
it's
kind
of
difficult.
It's
not
like
State,
it's
not
like
time.
It's
you
have
to
kind
of
predict
when
these
things
are
happening
and
they're
unevenly
distributed,
they're
like
earthquakes,
and
so
that's
those
are
the
four
kinds
of
the
way
differential
equation
we
can
have
now.
One
other
thing
to
remember
is
that
the
delay
function,
which
is
this
Tau.
A
What
you're
trying
to
do
overall
in
the
big
picture,
is
you're
trying
to
provide
a
solution
for
State
at
prior
times.
So
a
delay
generally
is
like
this
delay
of
you
know
kind
of
like
a
difference
between
the
curves,
but
it's
also
a
prior
time.
So
a
lot
of
times
you're
trying
to
solve
for
something
you
know
you
give
it
it
away.
You
say
something
like
this
system,
you
know
had
the
state
back,
you
know
several
moments
ago.
A
The
delay
actually
is
sort
of
the
mimic
of
that
state
down
the
line
so
you're
trying
to
use
like
past
Dynamics
to
sort
of
predict
things
later
that
that's
one
way
you
can
view
DeLay
So
you're,
using
solutions
from
prior
times
to
sort
of
approximate
what
the
current
state
is,
and
so
there
are
a
lot
of
different
ways.
You
can
use
the
way.
Dynamical
equations
I
hope
this
was
helpful.
It
looks
like
we
have
Quran
hello,
Quran.
B
A
Right
not
really
it's
cutting
in
now.
B
A
Yeah
so
yeah,
you
provide
an
update
in
the
chat.
That's
fine,
while
you're
doing
that.
I'll
mention
that
we
got
all
the
people
who
all
the
gsoc
students.
We
have.
Four
gsoc
students
and
everyone
turned
in
their
projects
and
I
received
a
notification,
and
then
I
did
a
sort
of
a
final
evaluation
for
everyone.
So
you
should
be
getting
those
next
week
and
everyone
passed.
A
So
that's
not
a
problem
that
looked
pretty
good.
Everyone
did
I
think
a
really
good
job
over
the
course
of
a
summer.
You
know
you
had
12
weeks
and
for
our
projects.
They
were
the
shorter
ones.
So
you
know
they
weren't
really
I
mean
I.
Guess
you
could
have
extended
the
deadline,
but
no
one
did
so
not
in
this
group.
A
B
A
Sure
and
they
they
worked
on
the
Grand
neural
networks
project,
so
the
dgnns
is
we
have
the
shorthand
for
so
that's
a
good
and
everyone
passed
and
I
was
pretty
pleased
to
see
what
was
turned
in.
So
that's
good.
So
now
we
have
these
two
sets
of
models.
We
have
the
digital
microspheres
and
the
graph
neural
networks,
so
I
I'll
just
give
a
little
brief
overview
of
what
we
have
here
with
the
graph
neural
networks.
A
This
is
the
dgnans
repo,
which
is
where
actually
that's,
not
the
right.
Well,
we
have
two.
This
is
the
old
one.
This
is
the
new
one
here,
devograph
I
might
just
rename
this.
This
says
nothing
in
it.
This
is
old,
I
put
I
need
this
and
I
forgot
and
then
I
moved
it
over.
So
this
is
Devo
graph
here.
A
This
is
the
sort
of
the
project,
so
this
was
all
being
done
on
Via,
Hong's,
repo
or
his
his
GitHub
account,
and
he
was
doing
it
there
and
coordinating
some
of
the
work
because
he
just
wanted
to
put
it
in
a
certain
place.
So
I
had
cloned
that
repo
and
brought
it
over
here.
So
this
is
all
basically
that
repo
with
the
readme
here,
basically
going
over
Devo
graph
introduction
to
this
I'll
need
to
rewrite
some
of
this.
A
For,
like
a
longer
term,
you
know
so
it's
presentable
longer
term.
We
have
Jiang
wataru
we're
the
students
we.
B
A
Wang
Lee,
who
was
an
external
collaborator.
He
did
some
work
on
on
one
of
the
part.
One
part
of
the
repo
Jesse
and
myself
were
mentors
so
yeah
we
had.
Then
we
have
the
list
of
contributions
here
and
yeah.
So
this
is
all
kind
of
like
a
listed.
What
Terrors
are
on
a
different
repo,
but
that's
okay
and
then
yeah.
So
we
have
the
the
program
in
here
Devo
graph,
that's
the
main
module.
A
We
have
these
different
stages
where
you're
bringing
in
data.
So
we
have
a
lot
of
test
data
that
are
available
in
the
repo
and
I.
Don't
know
why
it's
in
it's
in
Excel
format,
it
should
be
I.
Can
we
can
reformat
it
into
CSV
pretty
easily,
but
it's
actually
already
reformatted,
but
we
usually
work
with
CSV
files,
okay,
but
yeah.
A
So
this
is,
these
are
CSV,
so
some
of
them
yeah,
I,
guess
there
were
some
that
works
though,
but
they're
they're,
okay,
so
you
know,
CSV
files
are
just
common
to
limited
files
where
you
have
the
tabular
data
come
in
as
a
table,
they
populate
into
the
program
and
then
they're
analyzed
column
by
column
or
row
by
row
or
whatever
have
you,
and
so
we
have
some
data
that
we,
you
know
you
can
use.
A
We
have
the
main
module
here
and
yeah
I
think
it's
ready
to
test
out
that
people
are
interested
in
looking
at
it.
The
plan
here
is
to
take
devograph
or
dgnns,
and
you
know
combine
it
with
Divo
learn,
which
is
this
repository,
so
this
Repository
has
the
diva
learned
software
in
it,
and
I've
been
pretty
impressed
by
especially
wataru,
who
made
a
number
of
pull
requests
to
sort
of
make
a
second
generation.
Also
Holland
did
some
pull
requests
to
improve
upon.
A
You
know
what
was
there
in
the
repo
and
yeah.
So
this
is.
This
is
the
this
is
what
we
had.
This
was
two
years
of
Summer
of
code
and
some
other
contributors
sort
of
pulling.
This
along
and
now
we
have
this
Evo
graph
or
dgnns
or
whatever
you
want
to
call
them,
and
those
two
things
need
to
be
combined,
I,
guess
into
a
single
package
or
something
where
you
know
you.
Maybe
you
have
two
packages,
but
you
have
interoperability
between
them.
A
So
that's
the
goal
here
and
yeah
so
and
then,
of
course,
we
have
the
digital
microspheres,
which
is
sort
of
part
of
this
I.
Don't
know
if
it's
part
of
I
guess
it
could
be
part
of
Devo
Divo
learn,
but
it's
a
little
bit.
It's
a
little
bit
different.
A
We
have
a
devil,
worm,
AI
repo,
and
that
may
be
where
that
ends
up
living
I'm,
just
trying
to
figure
out
where
to
put
it
in
in
terms
of
sort
of
the
presentation
of
it.
You
know
how,
if,
if
I
were
to
send
someone
to
this
free
phone,
so
here's
what
we're
going
to
do
here
are
the
pieces
of
software.
You
need
to
do
to
have
like
a
really
nice
collection
tools
for
looking
at
development.
A
We
have
the
pre-trained
model.
We
have
the
graphql
networks
model.
We
have
maybe
some
modules
for
network
analysis
as
well,
which
are
separate
from
that
and
then
the
digital
microsphere,
so
you're,
taking
these
microscopy
images
and
putting
them
on
a
sphere.
That's
that's
the
idea,
and
so
let's
see
Quran
gave
an
update
all
right,
so
I'm,
still
working
on
improving
documentation
for
the
project.
This
week
is
mostly
spent
debugging.
A
B
A
Let's,
let's
keep
working
on
that
and
then
of
course,
especially
with
the
especially
with
the
digital
microspheres,
when
we
get
that
to
a
good
point
where
we
can
move
that
to
the
Divo,
learn
rebook
or
the
diva
learn
organization
will
do
that
we'll
create
a
repo
for
it.
A
But
you
know
this
is
kind
of
like
where
we're
pulling
these
things
together
and
trying
to
get
them
in
one
place,
and
you
know
getting
a
collection
tools
together.
So
that's
sort
of
the
road
map
in
verbal
form.
So
it
looks.
B
A
Susan's
here
and
yeah
yeah
she's,
the
only
new
person
since
I
last
checked.
B
A
A
B
A
A
Yeah
yeah
all
right.
Thanks
for
the
update,
so
yeah
I
guess
I
can
go
over
some
things.
We
had
I
have
a
couple
of
papers,
I'd
like
to
talk
about
and
new,
see,
elegant
stuff,
there's
a
couple
of
things
in
C
elegans
world.
That's
really
interesting!
That
have
come
out
recently.
A
A
If
you
want
to
create
a
CL
against,
if
you
want
to
simula
to
C
elegans,
there's
a
lot
of
data
out
there
behavioral
data
physiological
data,
there
really
are
a
part
of
open
worm,
at
least
not
yet,
and
so
one
of
the
things
that's
really
interesting
about
C
elegans
is
you
can
use
it
as
an
experimental
Tool
for
looking
at
heritability
and
evolution,
and
so
because
C
elegans
has
a
generation
time
of
about
three
or
four
days.
A
You
can
do
this
sort
of
thing,
so
you
take
a
c
elegans
egg
and
you
put
it
in
a
plate
and
you
can
let
it
hatch
and
expand
the
population
because
they
actually
are
hermaphrodites
mostly,
so
they
self
they
self-reproduce
and
they
end
up
putting
you
know
creating.
Maybe
in
four
days,
if
you
start
from
an
egg,
you
can
have
a
new
generation
of
c
elegans
and
that
generation
could
be
up
to
200
individuals,
because
that
one
worm
will
lay
their
eggs
and
those
eggs
will
hatch
and
they
will
become
CL
again.
A
So
you
have
basically
one
generation
every
three
or
four
days,
and
you
have
several
hundred
individuals
in
that
new
generation.
So
you
can
do
a
lot
of
nice
experiments
like
that.
So
this
paper
is
about
transgenerational
inheritance,
so
transgenerational
inheritance
is
where
you
have
the
inheritance
of
different
traits,
and
you
know
you
think
in
evolution,
you
think
about
like
Gene
frequencies
or
genes
that
get
inherited
or
genes
that
get
selected
for
in
trans.
A
What
they
mean
by
transgenerational
inheritance
here
are
largely
mechanisms
that
allow
for
short-term
generational
inheritance
so
say,
for
example,
you
know,
there's
like
something
like
an
environmental
stressor
that
environmental
stressor
is.
Can
you
know
information
about
that
can
be
inherited
to
the
next
few
Generations
ahead
of
you
reason
being
is
say:
if
you
have,
if
there's
a
scarcity
of
food,
you
want
to
have
protective
mechanisms
in
place
for
those
new
generations.
A
So
that
they
don't,
you
know,
starve
to
death
that
they
have
some
sort
of
set
of
sort
of
cues
that
are
already
there.
They
don't
have
to
learn
anything
new
to
to
survive,
so
this
is
and
and
what
they've
done
in
C
elegans
is
because
you
can
use
this
as
a
experimental
tool
for
evolution.
A
You
can
also
look
at
transgenerational
inheritance
pretty
nicely,
and
you
can
look
at
like
you
know,
maybe
like
you,
could
look
ahead,
10
or
20
generations
and
see
what
kinds
of
things
are
being
inherited
using
this
mechanism.
A
So
this
title
of
this
paper
is
transgenerational
inheritance
of
sexual
attractiveness
by
a
small
rnas
enhance
evolvability
in
C
elegans,
and
so
you
have
this
idea
of
evolvability,
meaning
that
you
know
you
can
evolve
in
different
ways.
So
evolvability
just
simply
means
you
know
being
having
the
ability
to
evolve
as
opposed
to
not
evolve.
So
just
basically
it's
this
ability
to
find
new
sort
of
phenotypes.
A
You
know
find
new
areas
of
the
of
the
phenotype
space
as
opposed
to
being
stuck
in
one
place
in
that
space,
and
so
this
is
what
they're
saying
here
is
that
these
small
RNA
molecules,
which
are
they
talk
about
in
the
paper?
These
are
typically
things
that
are,
you
know,
expressed
by
genes
that
are
present
in
the
cytoplasm
of
the
cell,
and
they
do
things
they.
Maybe
they
interfere
with
certain.
A
You
know
they
bind
with
certain
things
or
they
enhance
other
processes
that
you
know
there
are
different
ways
that
these
small
rnas
are
functional
and
it's
inheritance
of
sexual
attractiveness.
So
there
is
a
small
number
of
C
elegans
who
are
males,
and
you
know
most,
but
most
of
them
are
hermaphrodites.
So
this
is
one
of
the
things
they
do.
Here
is
a
graphical
abstract.
They
have
so
they
have
growth
at
25c,
which
is
growing
these
worms
that
typical
normal
temperature
for
them
standard
growth
conditions
20c.
A
So
you
have
this
in
this
case
you
have
20c,
and
then
you
have
a
little
bit
of
an
elevated
temperature
here,
usually
I
guess
they
grow
them
at
25c.
But
in
this
case
the
standard
Earth
condition
is
20c,
so
they're
raising
the
temperature
a
little
bit
in
this
case,
so
they
have
a
control,
condition
and
an
experimental
condition.
A
And
so,
when
you
have
this,
you
have
this
growth
of
25c.
Then
you
have
these
different
F1
f2f3
and
these
are
usually
Generations,
but
this
is
the
way
they
do
in
genetics
experiments.
They
say
that
these
F1
F2
F3
are
different
Generations,
where
they're,
selecting
and
back
Crossing
individuals,
so
that
they're
doing
these
experiments.
Where
they're
you
know
controlling
the
breeding
controlling
the
reproduction
and
you
look
at
The
Offspring
in
these
different
Generations
F1
F2
F3.
These
are
just
like
you
know
the
first,
the
second
and
third
generation
after
you
make
this
intervention.
A
A
You
know
to
start
with,
and
then
you
do
these
genetic
experiments
where
you
selectively
breed
different
worms,
and
so
that
means
maybe
taking
a
male
and
the
male
will
breed
mate
with
the
hermaphrodite
and
they'll
produce
Offspring,
and
so
that
ensures
that
this
transgenerational
inheritance
occurs
because
both
of
these
organisms
that
are
in
the
in
the
environment
that
you
put
them
in
have
been
exposed
to
the
stimulus
so
they're
going
to
inherit
this
mechanism.
A
And
so
then
you
have
this
transgenerational
inheritance
which
they
demonstrate,
and
this
is
inherited
attractiveness,
meaning
that
these
worms
somehow
are
more
attractive
to
mate
with
that
have
been
exposed
to
this
higher
temperature.
So
that's
what
this
paper
is
all
about,
so
the
in
brief,
which
is
the
abstract.
It
is
unclear
whether
transient,
heritable
epigenetic
responses,
so
transient
just
means
over
a
few
Generations
that
it's
it's
it's
apparent,
so
it
doesn't
last
forever.
A
It
only
goes
from
maybe
10
generations
and
then
that
effect
wears
off
and
you
have
to
to
be
re-exposed
to
this
higher
temperature
to
get
the
effect.
So
that's
what
they
mean
by
transient
heritable
epigenetic
responses.
The
small
molecule
is
something
that
is
trans,
you
know
transmitted
through
or
this
ability
I
guess
is
transmitted
through
these.
You
know
maybe
five
or
ten
generations,
and
then
you
know,
but
this
is
something
that
isn't
necessarily
like,
doesn't
rely
on
the
conventional
mechanisms
of
genetic
inheritance.
A
So
the
highlights
here,
if
you
grow
these
worms
at
25c,
if
you
expose
them
to
25c
and
then
even
if
you
bring
them
back
to
20c,
which
is
the
typical
set
of
growth
conditions,
you
can
increase
the
sexual
attractiveness
of
hermaphrodites,
increased
attractiveness,
transmits
transgenerationally
by
hrde1,
which
is
this
complex
that
they
talk
about.
They
have
a
picture
of
here
in
the
purple
and
small
rnas.
So
there's
this
this
mechanism
in
small
rnas
and
this
complex
that
are
responsible
for
this
transgenerational
effect.
A
Attractiveness
is
associated
with
sperm
defects
and
small
rnas
targeting
sperm
genes,
so
this
is
what
they
mean
by
you
know.
Attractiveness
and
C
elegans
is
quite
a
bit
different
from
like
what
it
might
be
in
humans.
It
could
be
that
they
have
a
you
know
that
they
they
have
a
well
I,
don't
know
what
it
means
exactly,
but
anyways
it's
associated
with
these
features,
sperm
defects
and
small
RNA
targeting
spread
genes.
A
So
these
are
different
things
that
you
know
are
that
allow
for
this
sort
of
enhanced
attractiveness
to
to
be
expressed
as
a
increased
reproduction
or
decreased
reproduction
stress
in
this
transient.
Epigenetic
inheritance
enhances
mating
and
can
affect
evolution
and
so
yeah.
So
they
talk
about
this
paper.
They
go
through
the
results
and
they
kind
of
talk
about
some
of
the
like.
They
always
look
at
mutant,
phenotypes,
so
hrd,
one
mutants
they
can
actually
in
C
elegans.
A
They
can
create
mutants
where
they
target
a
certain
Gene
and
you
know,
get
rid
of
that
Gene
and
then
see
what
the
effect
is
on
the
phenotype.
So
there
are
all
sorts
of
defined
mutants
that
you
can
get
for
C
elegans.
A
So
because
of
that
they
used
crispr
cast
9,
which
is
a
genetic
engineering
technique.
The
engineer
strains
permitted
conditional
hrd
one
depletion:
the
approach
allowed
for
the
depletion
of
hrd
one
protein
down
to
an
undetectable
level
within
one
hour
of
exposure
to
auxin,
which
is
a
chemical
stimulus.
A
We
validated
that
the
aux
independent
degradation
of
this
indeed
abolishes
a
transgenerational
RNA
eye
inheritance
by
examining
its
effect
on
the
heritable,
silencing
of
a
germline
expressed
transgene.
So
basically
they
wanted
to
try.
They
wanted
to
test
the
sort
of
the
efficacy
of
this
mechanism,
so
they
looked
at
mutants.
They
were
able
to
shut.
They
were
able
to
allow
for
depletion
of
this
Factor
using
a
ex
using
an
environmental
stimulus,
and
then
they
could
actually
so
they
do
a
lot
of
this
in
these
experiments,
where
they
test
different
things
they
they.
You
know.
A
The
yeah
what
they
do
with
crispr
is
they
can
Target
specific
locations
in
the
genome
and
they
can
take
out
a
gene
or
they
can
modify
a
gene
in
some
way,
and
so
they
could
basically
control
how
it's
expressed
or
if
it's
even
there
they
can.
They
can
remove
the
gene
itself.
B
A
So
the
thing
is
is
like:
when
you
have
a
what
what
they
call
Wild
type,
which
it
doesn't
have
this
mutation,
it
can
show
this
effect
of
enhanced
reproducibility
or
reproduction.
But
when
you
have
these
mutants,
which
don't
have
the
gene
they're
sterile
at
25
C,
so
this
enhanced
temperature
either.
You
know
if
they
have
a
mutation
or
it
makes
them
sterile
or
if
they
don't
have
a
mutation.
It
makes
them
sort
of
enhanced
in
terms
of
their
reproduction.
A
That's
what
they're
doing,
and
so
you
know
so
they
do
a
lot
of
these
type
of
experiments
just
to
Target
the
mechanism.
A
And
then,
of
course,
you
know,
you
see
the
effects
here
where
they
have
sexual
attraction
through
chemotaxis,
so
sexual
attraction
and
C
elegans
is
largely
because
C
elegans
usually
uses
a
lot
of
chemical
communication
and
chemical
stimuli.
So
they
have
this
chemotaxis,
it's
very
important
in
C
elegans,
and
then
they
can
show
that
you
know
there's
an
enhanced
sexual
attraction
at
25c.
This
is
the
the
what
I
talked
about
is
the
one
The
Knockout,
the
mutant.
A
It
doesn't
have
that
Gene
or
you
know
it
doesn't
have
what
they
need
to
show
this
effect.
This
is
another
mutant
oil
type,
which
also
has
high
sexual
attraction,
but
it's
a
different
Gene,
so
they
just
showed
us
as
opposed
to
the
20,
the
wild
type
raised
at
20
degrees,
Celsius,
and
so
they
do
a
mating
assay
and
they
they
show
how
you
do
this
over
Generations.
So
you
know
you
have
F1
through
F6
and
F
the
F's
just
refer
to
like
controlled
breeding
experiments
that
you
do
on.
A
A
What
they
call
you
know
back
Crossing
experiment
where
they
actually
want
to
take
like
a
mutant
and
isolate
the
mutant
and
make
sure
that
the
mutation
is
always
expressed
in
a
line
versus
experimental
Evolution
experiment
where
you
might
select
individuals
at
the
highest
Fitness
and
pass
them
the
Next
Generation.
So
this
is
a
nice
way
that
they
did
this
experiment.
A
But
the
bottom
line
for
this
paper
is
that
you
know
you
have
this
transgenerational
effect
and
that
it's
something
that
you
can
see
that
we
don't
really
model
in
our
models
of
sealed
and
so
I.
B
B
And
it
said
that
they
they're,
like
10
to
15,
like
with
with
the
different
conditions.
A
Oh
this
one
here.
A
Yeah
here
so
yeah
10
to
15
Generations.
Well,
this
is
before
the
F1.
This
is
before
the
back
Crossing.
So
they
do
this
like
they'll,
raise
it
at
that
many
generations,
so
I
guess
yeah.
The
effect
is
about
10
to
15
Generations,
so
you're
raising
it
at
this
elevated
temperature
and
then
you're
doing
this
back
Crossing
to
see
what's
responsible
for
it,
but
the
effect
yeah
usually
does
last
10
to
15
Generations
I'm,
just
not
putting
my
fingers
on
word.
A
That
says
it
where
it
says
that
in
the
paper,
but
so
that's
I
mean
that's
a
lot.
Okay,
I,
don't
know
yeah,
that's
a
good
amount
of
generations
of
C
elegans.
So
that's
it's!
It's
an
interesting
finding
and
it's
something
that
you
know
we
don't
in
open
worm.
We
don't
really
have
like
generational
changes,
it's
just
kind
of
like
a
model
of
the
canonical
worm,
and
so
it's
kind
of
interesting
to
see
how
that
works.
A
The
other
paper
I
was
going
to
talk
about
with
respect
to
new
C
elegant
stuff.
Is
this
paper
on
gonad
morphogenesis?
A
So
this
is
turning
back
to
the
stuff
that
we
were
doing
with
been
talking
about
Push,
Pull
morphogenesis
and
all
this
morphogenesis
having
to
do
with
physics,
the
physics
of
the
embryo,
and
so
this
paper
talks
about
the
C
elegans.
You
have
the
gonads,
which
are
the
reproductive
organs
and
how
morphogenesis
proceeds
there
and
the
title
is
direct
directed
cell
invasion
in
asymmetric
adhesion,
Drive
tissue
elongation
and
turning
in
C
elegans
going
in
morphogenesis.
A
So
that's
a
lot
of
words.
This
is
the
graphical
abstract,
though
you
have
this.
You
have
the
morphology
of
the
gonad
here
and
then
this
is
where
these
L2
L3
and
L4.
These
are
the
larval
stages.
A
So
this
is
after
it's
been
hatched,
but
before
it's
an
adult,
and
so
it's
still
developing
outside
the
egg,
and
so
you
can
see
that
the
membrane
is
kind
of
moving
out
here
at
the
end,
there's
forward
movement
as
the
gonad
elongates
and
then
there's
rotation,
which
is
where
the
gonad
turns
and
then
there's
this
u-shaped
gonad,
where
there's
forward
movement
at
the
end
of
the
gonad.
So
this
gonad
is
moving
outward.
The
distal
tip
is
moving
outward
and
curving
around,
and
so
the
the
worm
isn't
really
fertile
until
it's
an
adult.
A
So
this
is
all
pre
pre.
You
know
breeding
era
for
the
the
worm.
Like
you
know,
the
worm
is
only
putting
eggs
out
during
adulthood
before
that.
This
and
after
hatch
has
gone
at,
is
being
constructed.
There's
there's
morphogenesis,
and
so
you
can
see
that
there's
this
distal
tips
out
set
of
distal
tip
cells
at
the
distal
end
of
the
gonad
it
moves
out
and
then
it
curves
around-
and
this
is
basically
like
a
rotation
and
then
this
folding
over
of
the
gonad.
So
there's
a
paper.
A
You
know
these
are
all
these
papers,
that
kind
of
profile
with
their
papers
that
I'm
showing
you
today,
because
it's
easier
to
understand.
Arbor
wall
at
all
was
the
paper
that
x-ray
talks
about
this
in
more
depth,
investigate
the
mechanics
of
C
elegans
gonadomorphogenesis.
In
this
study
they
show
that
the
tissue
elongation
that
you
see
here
results
from
the
weeder
cell
Invasion
powered
by
localized
Matrix
degradation,
so
there's
some
set
of
leader
cells
that
are
responsible
for
this
morphogenesis.
A
They
move
out
in
different
directions
and
then
the
rest
of
the
rest
of
that
organ
will
follow
that
need.
So
you
have
leader
cell
Invasion,
which
is
just
invading
a
new
space,
a
space
and
moving.
You
know
accordingly,
and
this
is
powered
by
localized,
Matrix
degradation,
so
there's
a
degradation
sort
of
at
the
end
of
The
Matrix.
A
Here
the
cells
move
out
in
the
invaded
new
space,
and
then
they
form
this
elongation
and
turning
so
they
can
make
these
movements
that
are
coordinated
as
a
tissue
by
only
a
couple
of
cells
that
make
the
sort
of
the
leader
move
and
then
there's
degradation
of
the
old
Matrix.
So
that
there's
you
know
it
allows
this
change
to
occur
because,
as
you
notice
at
the
other
end
of
the
gonad,
there's
no
real
change.
A
It's
all,
basically,
the
same
morphology
so
well
asymmetric,
so
extracellular
Matrix
adhesion,
which
is
where
the
cell
is
adhering
to
The,
extracellular,
Matrix
or
the
space
between
the
cells.
So
you
can
see
all
these
hexagons
and
that's
a
simplified
view
of
it.
A
What
actually
happens
is
there's
a
lot
of
space
between
these
hexagon
boundaries
and
that
space
in
between
gives
us
extracellular
Matrix
and
there's
an
adhesion
there
that
creates
torque
that
drives
the
u-turning
of
the
gonad,
so
these
cells
are
kind
of
in
the
same
Matrix
there's
extracellular
Matrix,
which
is
usually
some
sort
of
material.
It's
it's
a
soft
material,
but
it
has
some
ability
to
generate
friction
if
you
move
against
it.
A
So
these
are
things
you
know
we
talk
about
these
in
something
like
the
gonad.
These
are
actually
also
applicable
to
to
our
diatom
examples
too,
because
the
diatoms,
although
there's
sort
of
single
cells
in
a
colony,
they
do
have
this
extra
cellular
matrix,
oh
just
immediately
outside
the
cell,
and
sometimes
they
generate
different
structures,
just
the
media,
just
adjacent
to
the
cell
or
in
between
two
cells
sliding
together.
So
this
is
something
that
is
I.
Don't
know
it's
it's
fairly,
well
understood
the
physics
of
it,
but
the
actual
effects
on
movement
and
Collective
movement.
A
This
is
something
I
think
that's
people
are
really
starting
to
figure
out.
So
the
highlights
are
the
distal
tip
cell,
which
is
a
specific
cell
type,
and
the
CL
sell
against
from
Aphrodite
gonad
is
pushed
by
a
proliferating,
germ
cells.
So
this
leader
cell
is
pushed
by
germ
cells
that
are
dividing
here.
A
So
the
germ
cells
are
dividing
they're
kind
of
forcing
these
leader
cells
out
this
Matrix
around
the
N
breaks
down,
and
then
the
forces
in
The
extracellular
Matrix
results
in
the
shifting,
because
there's
torque
being
generated
and
then
they're
pushed
in
this
direction.
So
it
forms
a
u-shape
or
it
makes
a
u-turn,
I
guess
extracellular
Matrix
degradation
by
the
DTC,
which
is
you
know:
oh
distal,
tip
cell
directs,
gonad
elongation,
distal,
tip
cell
extracellular
Matrix.
Adhesion
polarity
drives
gonadal
turning,
so
this
polarity
between
the
two
and
then
cdc42,
which
is
an
SRC.
A
These
are
different
molecules
and
this
in
this
process
regulate
distal,
tip
cell
integrine,
polarity
and
integrine
being
another
molecule
during
the
turn.
So
you
can
see.
There's
a
lot
of
molecular
sort
of
activity
going
on
here
and
there's
a
lot
of
physics
as
well.
So
that's
this
paper
and
again
this
paper,
just
kind
of
goes
through
some
of
this
work.
It's
very
you
know
deep
into
sort
of
the
cell
biology
of
it,
so
they
have
these
different
stains
that
they
show
the
gonad
and
it's
you
know
in
its
adult
morphology.
A
Here
you
have
some.
You
know
some
where
they're
treated
with
rnai,
so
they
knock
down
certain
Factor
molecular
factors
and
they
show
defects.
A
So
you
can
get
a
sense
of
what
molecular
factors
are:
are
responsible
for
a
proper
morphogenesis
as
opposed
to
defectual
morphogenesis,
and
so
they
kind
of
work
this
all
out
here.
This
is
a
nice
figure
of
I.
Don't
know
if
this
is
what
this
is
a.
This
is
a
simulation,
so
this
is
where
they
actually
simulate
the
gonad
and
they
show
this
physical
process.
So
there
are
no
molecules
involved
in
this.
This
is
pure
physics,
just
showing
the
sort
of
the
time
that
takes
for
proliferation
to
be
responsible
versus
autonomous
movement.
A
So
you
can
see
when
there's
no
proliferation
of
stem
cells.
You
get
this
positive
stress,
it's
an
autonomous
process,
it's
just
this
distal
tip
movement
and
it's
positive
stress,
which
is
the
physical
thing
driving
it,
and
you
can
see
that
it
doesn't
make
this
turn
completely.
It
just
kind
of
Curves,
and
then
you
know,
that's
it
if
you
have
proliferation
driven
DTC
movement,
so
the
distal
tip
is
moved
by
stem
cell
proliferation,
Behind
These,
leader
cells,
you
get
Negative
stress
and
the
negative
stress
actually
results
in
this
U-turn.
A
So,
that's!
That's
all
that
bought
that
paper.
So,
okay,
so
do
you
have
any
questions
about
those
two
papers?
I
just
wanted
to
give
people
a
taste
of
what's
new
in
C
elegans
and
give
you
an
idea
of
what
we're
not
simulating
necessarily.
B
Both
from
the
technion
I
saw
like
the
the
lab
that
produced
the
produced
that
they're.
B
A
Interesting
yeah,
okay,
so
I
think
that's
it
for
this
week,
thanks
for
attending
and
Amman,
thank
you
for
the
update
on
the
mathematics.
I
thought
that
was
nice
yeah
I
keep
working
on
that
I.
Don't
know
you
know,
maybe
there's
some
insights
that
we
can
make
in
that
area.
I
know
that
Thomas
has
been
working
on
some
papers,
but
you
know
if,
if
there's
something
that
really
interests
you,
you
know
we
can
pursue
it
a
little
bit
further.
B
Yes,
yeah
good,
all
right,
all
right!
Okay,
thank
you,
guys.