►
From YouTube: DevoWorm #26: GSoC project updates, Systems Biology Integration, Push-pull Pattern Formation II
Description
GSoC Coding period (week 7): Updates on Digital Microspheres and D-GNNs. Mathematics of DevoWorm poster. Discussions of papers on unifying methods in Systems Biology, and a second set of papers on the physics of pattern formation (push-pull dynamics) in chick embryo and cell culture. Attendees: Susan Crawford-Young, Harikrishna Pillai, Longhui Jiang, Karan Lohaan, Morgan Hough, and Bradly Alicea
A
B
A
B
C
B
C
B
I
need
reading
glasses,
anyways
yeah,
so
I've
been
taking
it
easy
sort
of
yeah
for
two
weeks.
It'll
be.
If
I
do
this
for
three
weeks,
then
things
will
have
recovered
enough,
that
I
should
be
back
on
board.
A
Yeah,
that's
good,
as
I
guess
it's
the
summer,
so
you
get
like
they're
lower
expectations
for
your
graduate
program.
Oh.
B
Absolutely
and
I'm
not
doing
it,
so
it's
not
not
really
a
good
thing.
B
Yeah
I
found
a
few
papers
that
were
very
interesting.
I
could
pass
them
on
to
you.
I
guess
yeah
it'd
be
good.
C
B
C
D
D
Other
than
that,
I
think
this
for
week,
seven
from
july
25th
till
31st-
it's
mostly
just
categorizing,
whatever
we've
done,
you
know
putting
it
properly
under
a
template
with
you
know
proper
documentation.
So
this
is
what
we
planned
for
this
week.
D
D
The
major
thing
that's
going
on.
Otherwise,
apart
from
the
projection
method,
the
model
is
okay,
don't
done
too
many
changes
after
that,
yeah.
A
Looking
ahead,
is
there
anything
you
can
do
sort
of
as
you're
searching
for
the
better
method
I
mean?
Are
there
other
things
that
you're
kind
of
moving
toward
moving
forward
with.
D
D
So
the
thing
is
opencv
used
to
have
these
algorithms,
but,
since
you
know,
google
has
patented
them,
so
it's
kind
of
like
in
a
third
party
area,
packing
ways,
so
there
are
like
alternatives
for
this,
so
I'll,
currently
thinking
of
going
through
orb
as
another
alternative,
you
know
to
replace
this.
Otherwise
this
was
again
something
I
was
looking
into,
and
the
projection
method,
of
course,
is
another
thing
I'll
be
doing
this
week
as
well.
D
No,
it's
it's
like
it's
like
the
python
library
that
I
was
using
opencv
used
to
support
using
these
algorithms
with
the
library
itself.
So
you
know
I
didn't
have
to
hardcode,
or
you
know,
create
an
algorithm
for
each
specific
case
on
my
own.
They
used
to
have
you
know,
boilerplate
code,
that
used
to
do
it
for
you.
D
So
I
in
the
recent
versions
of
opencv,
this
thing
is
not
there
or
they've
removed
the
algorithm.
You
know
that
used
to
do
all
this
processing
on
their
own
end,
I'm
looking
into
alternatives
for
this.
Otherwise
I
can
use
shift
tensor.
It's
just,
I
think,
just
for
the
project's
sake.
You
know
I
was
looking
to
going
through
some
different
options,
because
the
opencv
libraries
dependencies,
you
know
that
will
be
included
in
the
file
project-
could
affect
somebody
in
the
future.
D
D
D
Shift
and
surf
because
I
think
it
was
something
regarding
the
licensing
of
you
know
open
source
projects,
so
they
can't
keep
it
open
source.
It's
it's.
I
don't
know.
The
reason
for
this
is
kind
of.
D
Like
even
I
don't
know
it
properly,
but
the
main
gist
of
it
was
that,
since
you
know,
the
engineers
at
google
had
come
up
with
the
idea
and
google
usually
gives
away
everything
for
open
source,
it's
just.
They.
A
D
What
would
the
subscripts
of
algorithms
that
are
there
and
you
know
them
not
being
there
in
the
newer
versions
of
opencv,
and
you
know
all
basically
placement
that
I'll
be
trying
out
within.
D
A
Okay,
yeah,
I
don't
know
I
I
don't
know
if
you
caught
my
question,
but
is
there
a
way
do
you
have
any
references
on
the
sifsoff
algorithm
that
we
can
like
a
wikipedia
page
or
something
that
we
can
review?
A
D
C
A
All
right,
thank
you,
hari
krishna,
do
you
have
an
update.
C
Yeah,
hello
yeah,
so
in
the
last
week
I
told
that
I
had
collected
those
3d
coordinates
in
a
file.
So
now
the
extension
is
called.ply.
C
C
Now
now
what
I
want
to
do
is
project
the
images
onto
the
mesh
which
I've
got
now
so
I'll
be
working
on
the
projection.
A
All
right,
that's
good!
You
were
having
some
technical
difficulties
before
like
with
trying
to
get
some
things
to
work.
C
A
D
A
Okay,
well,
that's
good
yeah,
so
this
week
we'll
be
hit.
Thank
you
for
the
update
yeah.
So
this
this
week,
I'm
gonna
be
doing
the
evaluations
for
gsoc
and
we
have
four
people.
We
have
krishna
koran,
otaru
and
jia.
Hong
and
I've
been
talking
with
watarun
jia,
hong
I've
been
keeping
updated
on
slack.
A
So
I
think
all
of
you
are
going
to
pass
and
I
just
need
to
write
up
the
evaluation
and
you'll
be
getting
copies
of
that
later,
but
I
mean,
like
you
know,
I
I
like
koran's
idea
of
writing
up
a
blog
post
on
that
an
update
you're
not
required
to
do
that,
but
it
would
be
nice
to
kind
of
reevaluate
your
project
at
every
evaluation
step,
just
to
make
sure
that
it's
still
within
scope.
That's
all!
I
was
asking
about
what
you
plan
to
do.
Looking
forward.
A
I
wanted
to
make
sure
that
we
were
keeping
on
with
the
schedule.
You
know,
and
we
have.
I
guess
one
of
the
things
they
asked
me
was
that
if
people
need
extensions,
we
can
request
them.
So
typically
the
program
runs,
I
think,
12
weeks,
and
then
you
know
you
get
your
evaluation,
but
if
you
want
to
extend
it,
if
you
need
to,
you
know
submit
it
later,
maybe
in
the
fall,
I
think
they
allow
you
to
extend
it
up
to
22
weeks.
So
that's
pretty
flexible,
hey.
A
I
can
request
an
extension
and
then
you
have
a
new
due
date.
So
I
don't
think
I
don't
know
if
anyone
in
this
group
will
be
requesting
an
extension.
We
have
one
of
the
shorter
projects,
so
the
longer
projects,
usually,
I
think,
are
going
to
get
maybe
get
the
extension
given
more
freely
and
we
have
in
my
other
group,
we
have
a
longer
project
time.
So
that's
I
don't
know
how
relevant
that
is.
But
if
you
do
need
an
extension,
please
let
me
know:
okay
just
wanted
to
go
over
actually
morgan's
here,
hello,
morgan.
A
Oh
thank
you
for
the
link
bro.
I
just.
A
D
From
morgan,
I
just
have
a
clarification
regarding
the
thing
so
yeah
it's
been
discontinued
in
opencv,
but
it's
not.
I
think
it's
not.
It
isn't
patented
by
its
old
creators
the
people
who
created
the
algorithm.
So
it's
there
for
free
and
academic
use
and
research
settings.
It's
it's
it's
free
for
that.
I
think,
but
the
person
will
be
using
the
library
he'll
face
some
issues.
I
think
that's
the
only
thing.
D
B
Yeah
hi
garland
yeah
not
much
of
an
update,
I'm
just
just
listening.
C
And
yeah
caught
your
embryology,
video,
which
yeah
still
still
processing
but
but
that
was
great
and
yeah
all.
A
Right,
yeah,
good
nice
to
hear
nice
to
have
you
be
able
to
make
this
meeting
so
yeah,
so
I
do
want
to
go
over
some
things
here.
The
first
one
is
this
article
that
quran
it's.
Let
me
share
my
screen
here.
A
A
Detection
algorithm
the
sift
and
surf
algorithms
are
patented
and
free
to
use
in
academic
settings,
and
then
so.
This
is
the
sift
algorithm
here,
and
this
has
the
code
is
here
and
then
there's
a
link
to
the
it
gives
it
some
documentation
on
it.
A
A
Local
features
are
the
key
and
you
have
detectors
and
descriptors
yeah.
So
this
is
gonna.
This
is
basically
the
structure
of
sift,
find
scale,
space,
extrema,
key
point,
localization
and
filtering
and
per
keep
improve
key
points
and
throw
out
the
bad
ones.
A
A
D
Yeah
yeah,
it's
it's
basically,
you
have
when
you
have.
You
know
two
different
views
of
an
object,
be
it
views
from
a
distance
two
different
distances
or
from
two
different
angles.
It
helps
you
in
you
know
finding
the
key
points
that
are
there
and
using
this
algorithm,
I'm
using
it,
for
you
know,
finding
the
angle
of
rotation
and
do
three
more
other
things.
You
know,
because
I
know
the
key
points
that
are
there,
how
much
they've
moved-
and
you
know
for
like
these-
are
the
two
three
functions.
D
You
know
that
I'll
be
using
that
I'm
using
for
generating
the
model,
the
amount
that
I
have
to
rotate
the
outlines
as
well.
You
know
these
sort
of
things,
so
in
that
blog
only
you
know
he's
mentioned
a
better
way
to
go
about
these
things,
because
these
things
take
time.
So
orb
is
currently
the
most
recent
algorithm.
I
think
for
doing
something
similar
so
shift.
Is
there
then
surf
is
there
after
that?
The
next
best
thing
is
all
so
I'll,
be
using
thinking
of
even
incorporating
that
as
well.
D
Otherwise,
the
main
useful
aspect
of
this
algorithm
was
the
key
point.
Detection
and
you
know
finding
the
displacement
of
points
and
all
those
things.
C
C
A
Of
alternate
version,
or
some
alternate
method
that
you're
going
to
use,
but
it's
going
to
basically
do
a
similar
thing.
D
A
A
So
this
is
where
you're
taking
this
fat
or
the
sift
model,
or
no
they're,
taking
the
fast
model
and
the
brief
model
when
they're
putting
it
in
turning
it
into
this
orb
model,
and
they
don't
have
a
document
here,
but
this
is
the
code,
so
it's
using
very
similar
things
key
points
descriptors
and
then
it's
turning
it
into
that
sort
of
projection.
A
So
that's
nice!
That's
interesting!
Thank
you
for
that.
Sharing
that
link
in
terms
of
pull
requests
on
diva
learn.
We
have
wataru's
been
making
some
pull
requests
here
for
different
things.
We
have.
A
You
know
we
haven't
done
a
lot
of
them
lately,
but
he's
made,
I
think,
one
other
pull
request
the
summer
code,
so
he's
working
on
the
gnns
project
and
he's
working
on
importing
data
and
getting
that
pipeline
set
up
so
he's
working
on
the
you
see
it
was
having
some
problems
with
the
cell
segmenter
and
he's
been
improving
upon
that.
So
congratulations
to
wataru.
For
that
also.
A
I've
talked
about
the
mathematics
of
diva
worm
in
past
meetings,
and
so
I
finally
got
this
up
and
like
into
a
poster
form
and
it's
going
to
be
presented
at
find
your
inner
modeler,
which
is
a
an
nsf
sponsored
workshop.
A
We've
we've
had
about
five
of
them
so
far,
so
this
is
the
fifth
one
and
it's
it's
a
nice
event.
It
brings
together
biologists
and
modelers
and
brings
them
together
in
one
conference
and
then
the
conference
setting
they
basically
try
to
find
different
ways
for
them
to
talk
to
one
another.
So
you
know
biologists
will
give
a
talk
on
their
biology.
The
modelers
will
give
a
talk
and
they're
modeling.
Sometimes
you
have
people
a
lot
of
people
with
both
sets
of
expertise.
A
So
you
know
they
they
give
talks
and
they
kind
of
put
the
biology
in
the
context
of
modeling.
So
this
is
the
mathematics
of
divaworm
poster
for
this
particular
event-
and
you
know,
I've
tried
to
take
some
of
the
parts
that
I've
shown
you
in
previous
meetings.
A
Let
me
see
if
I
can
get
this
at
a
better
size
here,
all
right
and
add
some
biological
context,
some
modeling
context,
and
then
this
is
what's
going
to
be
in
so
you
know,
basically
in
a
poster,
you
want
to
be
very
visual.
So
all
of
these
different
points
here
that
are
labeled
with
letters.
These
are
all
from
papers
that
we've
done
in
the
group
at
various
points
in
the
past.
A
So
you
know
we
have
these
different
things
that
you
know
we
want
to
do
things
in
to
characterize
development,
different
aspects
of
development,
and
we
have
some
models
that
we've
proposed
we've
written
about,
and
so
that's
what
this
poster
is
about.
It
just
kind
of
goes
through
those
different
models
for
different
things.
So
the
first
one
is:
how
do
we
capture
the
multitude
of
interactions
and
state
changes,
or
so
which
could
be
like
cell
identity
or
other
things
that
occur
across
embryogenesis
and,
in
this
case,
we're
thinking
of
discrete
dynamical
and
categorical
models?
A
This
one
is:
how
do
we
capture
pattern?
Formation
and
cell
migration
dynamics
is
a
computational
process
and
there
our
answer
is
cellular
automata
and
spatial
data
structures.
So
we
have
some
examples
here.
We
have
an
example
of
a
cellular
automata,
producing
a
sort
of
a
modeled
pattern
or
a
dotted
pattern,
and
so
this
is,
you
know,
pattern
formation
in
a
cellular
automata.
A
A
This
volume
can
also
be
extracted
from
some
of
the
work
that
we're
doing
on
projecting
to
a
sphere.
So
this
is
also
something
that
can
be,
and
then
the
angle
might
be
like
the
angle
at
which
it's
projected
to
I
mean
there
are
different
ways.
We
can
do
this,
but
this
is
the
the
sort
of
the
core
representation
here
and
then
then
there
are
these
other
things
like
well.
A
Actually,
this
is
a
different
question,
but
this
anyways
this
is
the
c,
and
then
this
one
is
d,
which
is
how
do
we
characterize
the
organization
of
developmental
cells,
expanding
more
for
morphological
structures
and
emerging
connectoms,
and
our
answer
is
neural
networks,
complex
networks
and
trees.
So,
every
time
we
get
asked
a
question
we
give
an
answer,
and-
and
the
answer
that's
in
bold-
is
our
group's
answer
to
that-
not
necessarily
like
the
definitive
answer,
but
it's
just
one
possible
answer.
A
A
So
this
if
we
go
to
the
diagrams
here,
this
includes
node
attachment
and
complex
networks.
So
this
is
where
we're
growing
networks
from
maybe
two
nodes
to
50
nodes
or
500
nodes,
and
we
have
to
add
a
nodes
every
time
you
know
we
have
some
growth
event.
So
if
something's
hap
growing,
if
it's
adding
nodes,
if
cells
are
dividing
or
if
our
scope
is
expanding
or
if
the
shape
is
changing,
then
we
add
nodes
and
we
have
to
have
some
sort
of
attachment
process
for
that.
A
We
also
work
with
neural
networks.
So
neural
networks
are
interesting
because
they're,
although
they're
different
types
of
networks,
so
they're
they're
not
like
complex
networks,
they
still
need
different
rules
to
have.
This
is
really
not
okay.
There
we
go,
they
still
need
different
rules
and
we
use
these
to
model.
The
nervous
system,
perhaps
of
a
worm
instead
of
the
morphology,
so
he's
in
a
different
place.
A
A
A
I
believe
this
is
a
computational
study
on
looking
at
what
happens
when
you
flip
the
nodes
at
a
particular
level
of
the
lineage
tree,
and
you
also,
if
you
can,
if
you
represent
it
as
a
so,
if
you
represent
it
as
a
differentiation
tree,
it
looks
different
than
if
you
represent
it
as
a
lineage
tree
in
c
elegans.
A
It's
really
about
ordering
the
nodes
from
smallest
to
largest,
instead
of
the
thinking
in
terms
of
pure
heredity.
So
you
have
two
alternate
trees.
You
have
a
differentiation
tree
and
you
have
lineage
tree,
then
what
you
can
do
is
you
can
create
random
conditions,
so
you
can
create
trees
with
nodes
that
are
just
jumbled
around.
A
Then
you
can
compare
those
two
and
see
what
their
information
is
and
there's
an
information
measure
on
the
paper
that
we
use,
and
you
see
that
in
c
elegans,
when
you
compare
like
biological
trees,
which
are
the
ones
that
have
some
ordering
criterion
to
random
trees,
that
there's
this
difference
in
terms
of
their
information
or
their
hamming
bees
hamming
distance
in
this
paper,
but
basically
there's
this
greater
amount
of
information
in
the
tree,
and
so
it's
kind
of
an
interesting
study.
A
But
I
you
know,
I
don't
want
to
get
any
more
into
it,
because
it's
one
of
these
things
that's
hard
to
it's
hard
to
describe
any
more
than
that
without
reading
the
paper.
A
So
this
is
an
example
of
a
lineage
tree,
and
this
shows
you
the
mapping
between
the
lineage
tree
and
the
the
number
of
cells
in
an
embryo.
So
this
is
from
someone's
paper
here.
I
just
use
this
as
a
an
example
of
a
lineage
tree,
but
then
this
paper
was
from
2021.
This
was
on
looking
at
cell,
simulating
cell
divisions
using
a
number
of
distributions
so
that
you
know
that
shows
that
certain
distributions
are,
you
know,
provide
provide
you
with
different
patterns
than
others
in
biological
cases.
A
It
looks
more
like
this
yellow
function
than
the
other
functions.
So
there's
you
know
it's
kind
of
following
up
on
this
work,
except
it's
in
a
little
bit
different
way,
but
anyways.
We
are
interested.
We've
been
interested
in
this
idea
of
lineage
tree
dynamics
and
the
question
is:
is
how
do
we
characterize
the
rate,
recombination
and
conservation
of
growth
and
form?
And
the
answer
is
lineage
and
differentiation
tree
modeling.
A
A
But
you
can
use
that
to
build
these
vehicles,
build
these
complex,
phenotypes
and
trace
out
the
differentiation
tree
so
you're
using
different
types
of
what
we
you
know.
We
could
call
tissues
which
are
just
different
parts
of
the
of
the
vehicle
and
then
you're,
letting
it
differentiate
and
divide
into
new
parts
and
you're
tracing
all
that
through
one
of
these
linear
differentiation,
trees
and
then
yeah.
So
that's
the
end
and
then
at
the
end
we
have
these
references,
which
are
the
different
references
that
we're
using
in
our
group.
C
Oh
yeah,
how
are
you
hi
sorry.
C
A
It's
fine!
So
how
is
how's
the
graph
neural
networks
project
going.
C
Yeah
the
stage
ones
this
source
code
is
releasing
and
I
use
it
to
process
the
video
yeah
and
and
then
I
was
studying
the
results.
Yeah.
C
C
So
yeah
stage
one
is
going
yeah,
okay,
so
stage
two
is
work.
Yeah.
A
Good
yeah,
that's
good!
So
it's
good!
That's
all
making
progress!
It's
pretty
ambitious,
but
I
think
it's
doable.
We
can
get
it
done.
You
know
it
can
be
done
so
that
that's
good
and
I
saw
that
otaru
had
posted
some
pull
requests
for
divo
divalearn.
A
Okay.
So
that's
good
yeah!
Do
you
know
what
jiahung
and
wataro
do
you
know
if
they're
how
they're
making
progress
or.
C
C
Because
the
source,
the
source
code
is
great,
so
I
just
can
study
it
well.
Yeah,
okay,.
A
Well,
that's
good
yeah!
Thanks
for
the
update
and
I'll.
A
All
right,
thank
you.
Did
we
have
any
questions
about
anything
that
we
meant
talked
about
in
that
previous
section.
C
A
Well,
we
have
a
I
I
want
to
get
into
some
papers
now.
If,
if
you
don't
mind,
I
have
some
so
I
I
last
week
I
recorded
an
update
in
addition
to
people's
gsoc
updates,
and
I
put
it
into
the
video
as
on
youtube,
and
there
are
some
two
topics
that
were
covered.
One
is
a
really
interesting
set
of
papers
on
sort
of
an
alternate
alternative
to
reaction,
diffusion
or
morphogenesis,
and
the
other
was
on
critical
periods,
light
exposure
and
critical
periods
and
shifts,
which
is
interesting.
A
I
mean
it's,
usually
you
know
very
behavioral,
but
it's
also,
you
know
it's
part
of
development,
so
this
week
I'm
going
to
go
through
two
sets
of
papers,
so
the
first
one
is-
and
this
might
you
might
find
this
interesting
in
terms
of
systems
biology.
A
I
know
a
lot
of
you
are
maybe
familiar
with
systems
biology,
but
if
you're
going
to
be
doing
computational
modeling,
this
is
important
to
know
how
this
at
least
be
aware
of
this
kind
of
work.
So
systems
biology
is
really
an
area
that
you
know
where
they.
They
do
focus
a
lot
on
genomic
data
and
things
like
that.
A
But
still
you
know
there's
this
whole
field
of
systems
biology
people
are
also
interested
in
physiology
like
measuring
physiology
in
real
time
or
you
know
proteomics
or
things
like
that.
So
there's
it's
pretty
diverse,
so
this
is
papers
on
modeling,
formalisms
and
systems
biology,
and
this
is
out
of
amb
express
and
I'm
not
really
familiar
with
that
journal.
But
anyways
the
abstract
reads:
systems
biology
has
taken
advantage
of
computational
tools
and
high-throughput
experimental
data
to
model
several
biological
processes.
A
These
include
signaling,
which
is
like
chemical
signaling
between
cells
or
signaling,
within
cells,
gene
regulatory
and
metabolic
networks.
So
gene
regulatory
networks
are
these
networks,
that
of
genes
that
produce
some
sort
of
gene,
product
or
gene
expression.
So
mrna
or
you
know
even
their
role
in
producing
proteins,
but
you
know
in
an
indirect
way
and
then
metabolic
networks,
so
metabolic
networks
are
where
they
look
at
metabolites
and
they
look
at
other
types
of
or
they're
really
vast
networks
that
they
model,
and
you
know,
they're
chemical
processes
in
the
in
the
cell.
A
So
you
know
you
think
about
things
like
atp
and
other
types
of
you
know:
energy
production
and
other
types
of
things
like
that.
That's
what
they're
doing
with
that
gene
regulatory
networks
are,
of
course
you
know
they
give
you
some
readout
of
maybe
like
mrna,
and
you
can
look
at
the
fluctuations
both
in
you
know,
at
a
single
time
point
and
over
time
and
oftentimes
you'll
find
things
like.
You
know:
different
types
of
fluctuations,
different
types
of
rhythms,
and
so
it's
it's
interesting.
A
I
mean
most
people
just
measure
gene
expression
with
relation
to
like
an
experimental
condition
like
if
you
take
out
you
know,
if
you,
if
you
treat
it
with
a
drug
and
these
cells,
you
know,
respond
in
kind
and
you
take
your
measurement
of
gene
expression,
and
you
say
this
is
you
know
it's
upregulated
or
down
regulated
as
compared
to
something
that
didn't
experience
the
chemical
treatment?
A
But
you
can
also
look
at
gene
regulatory
networks
over
time
and
in
fact,
that's
where
I
think
a
lot
of
the
information
there
lies.
A
So
there
are
all
sorts
of
the
focus
on
networks,
but
you
know
systems
biology,
features,
a
lot
of
other
types
of
structures-
and
you
know
such
as,
like
you
know,
putting
the
data
into
an
atlas,
mapping
into
tissues
and
other
types
of
things
and
the
term
systems
is
very
loose.
You
know
we
talk
about
systems,
usually
we
talk
about
more
than
one
part,
so
it's
really
about
a
lot
of
parts
interacting,
so
you
can
use
networks,
but
you
can
use
other
things
as
well.
A
However,
in
in
the
field
of
systems
biology
most
of
the
models
that
they
use
are
specific
to
each
kind
of
network,
so
your
networks
are,
you
know
they
use
a
certain
type
of
network
for
a
certain
type
of
problem.
There
aren't
really
network
models
that
bridge
every
area
of
the
field.
A
A
A
I
can't
really
reconcile
those
networks,
it's
hard
to
like
bring
them
together
and
say
you
know
this
is
happening
in
gene
expression.
This
is
happening
in
in
metabolomics
and
here
is
the
connection
it's
very
hard
to
do
so
there,
but
these
systems
do
interconnect
and
really
that's
also
part
of
the
system's
approach,
but
because
of
the
way
these
systems
or
these
network
types
have
been
developed,
it's
very
hard
to
make
that
connection.
A
P
tree
nets
which
are
kind
of
mysterious.
To
me,
I
don't
I've,
never
understood
them,
but
it's
a
different
type,
very
different
type
of
network
modeling
approach,
process,
algebras
constraint,
based
models,
differential
equations,
rule-based
models,
interacting
state
machines,
cellular
automata
and
agent-based
models.
A
So
that's
a
pretty
broad,
swath
and
they're
going
to
review
all
these
methods
and,
like
you
know,
really
think
about
like
these
different
methods
and
they
have
they
sort
of
do
different
things.
So
you
know
you
have
your
differential
equations,
they're
modeling,
some
sort
of
time
series
they're
modeling
a
system,
that's
unfolding
in
time,
whereas
your
you
know,
agent-based
models
do
the
same
thing,
but
they
do
in
a
very
different
way.
They
do
it.
A
I
can
put
this
in
the
channel
on
the
slack
if
you're
interested
in
reading
more
about
it,
I'm
not
going
to
get
deeply
into
it,
but
I'm
just
going
to
go
over
through
here
and
see
if
there
are
any
nice
takeaways
to
bring
up.
So
you
know
they
get
into
this
idea
of
biological
networks.
A
So
here's
an
example,
a
signaling
network,
is
on
the
left,
where
you
have
this
sort
of
topology
and
you
have
these
symbols,
and
you
know
it's
they're
you're,
looking
at
a
very
specific
set
of
relationships,
you're
looking
at
signaling,
you
know
from
one
component
to
another,
then
you
go
to
gene
regulatory
networks
where
you
have
a
very
different
structure.
You
have
your
genes,
you
have
your
products,
you
have
your
promoters
and
your
enhancers
and
all
that
and
then
it's
producing
something
some
sort
of
product,
some
mrna
or
something
like
that.
A
And
then
you
have
metabolic
networks
which
are
these
metabolites
that
are
connected
together
and
there
are
different
rates
and
the
rate
limiting
factors,
and
you
know
so
there's
this
these
dynamics
inside
these
networks
that
are
really
interesting.
A
But
the
the
key
problem
here
is
that
you
have
a
difficulty
sort
of
reconciling
these
networks
so,
for
example,
this
green.
These
green
boxes
are
proteins
in
this
model
and
the
proteins
in
say,
like
a
signaling
model,
might
connect
to
regulation
of
a
gene
in
a
gene
regulatory
model.
And
so
we
can
map
that.
But
it's
very
hard
in
practice
to
actually
calibrate
the
model
with
with
data,
because
you
know
we
can
say,
follow
these
signaling
pathways
to
the
production
of
proteins
and
then
we
can
follow
it
to
a
general
regulatory
model.
But
the.
A
A
A
It's
it's
a
very
hard
thing
to
actually
integrate
these
models
so
they're
showing
an
example
where
you
know
this
is
a
very
generic
example
they're
showing
how
they
might
interconnect
a
metabolite
might
affect
the
expression
of
a
gene
as
well.
So
this
is
these.
Are
the
rna
squares
here
and
these
are
the
proteins,
so
it
was
mistaken
on
that.
But
basically
the
idea
is
that
things
can
regulate
a
gene.
You
have
them
coming
in
from
different
models
now
they're
very
hard.
A
These
models
are
very
hard
to
implement
and
calibrate
and
get
them
to
work
well
together.
But
you
can
draw
these
pictures
and
you
can
show
how
they
fit
together,
so
they
go
over
a
lot
of
they
actually
go
over
topological
analysis,
which
is
something
we
talked
about
in
the
gnns
group.
We've
talked
about
this
in
our
our
group
previously,
maybe
last
year.
I
think
we
discussed
this
a
bit.
A
So
people
are,
you
know,
there's
this
move
towards
topological
analysis.
So
there
are
a
lot
of
things
you
can
do
with
these
networks.
Once
you
establish
them,
you
can
look
at
the
things
like
modularity
and
hierarchy,
which
are
the
structures
that
occur
within
networks.
A
If
you
get
like,
you
know,
groups
of
things
that
group
together,
that's
modularity.
If
you
get
things
that
are
sort
of
organized
underneath
one
another,
that's
hierarchy,
there
are
different
ways
that
you
can
characterize
structure
with
networks,
but
another
thing
we
might
want
to
do
is
use
topological
analysis,
which
is
so
they
say.
A
considerable
amount
of
work
in
this
field
is
based
on
topological
analysis
of
biological
networks.
A
In
this
case,
graph
based
representations
also
play
a
fundamental
role.
The
analysis
of
the
topological
properties
of
these
graphs,
such
as
degree
distribution,
clustering
coefficient
shortest
paths
or
network
motifs,
can
reveal
crucial
information
from
biological
networks,
including
organization,
robustness
and
redundancy.
A
So
you
know
we
want
to
be
able
to
analyze
the
topology
of
the
network,
and
you
can
do
that
through
network
statistics
and
other
things
like
that,
but
you
can
also
actually
analyze
the
structure
of
the
network
and
the
structure
of
the
data,
in
fact
to
form
the
network,
and
so
that's
where
this
topological
analysis
comes
in,
but
there's
also
topological
data
analysis
which
can
be
used
to
inform
the
formation
of
these
networks
so
yeah,
so
that
they
really
stress
standardization.
A
So
if
you
are
familiar
with
the
systems
biology,
markup
language,
this
is
really
the
standard
of
the
community.
This
is
you
know
in
order
to
sort
of
make
models
that
people
can
use
across
the
community.
We
need
to
have
a
common
language,
so
one
of
the
ways
they've
done
this
is
to
work
on
the
system's
biology,
markup
language-
and
this
is
a
you
know-
a
standard
that
people
use
to
build
models
for
many
different
systems,
and
you
know
then
they
can
share
them
with
people.
Just
like
we
share
pieces
of
software.
A
We
can
say
these
are
the
specifications
and
you
know
you
can
just
plug
in
your
data
and
it
should,
you
know,
be
like
a
standardized
format,
so
these
models
up
here
are
standardized
in
the
sense
that
they
have.
You
know
we
have
symbols
for
different
things.
So
if
it's
a
receptor,
it's
a
symbol.
If
it's
a
signal,
signaling
molecule,
it's
this
purple,
symbol,
proteins
are
green
light.
Green,
mr
rna
is
dark.
A
Green
metabolites
are
this
sort
of
off
green
and
so
forth,
and
so
you
have
this
convention
where
you
can
have
a
legend
and
it
tells
you
what
these
things
are
and
then,
of
course,
if
it's
something
that's
common,
you
know
you
can
also
account
for
that
in
the
in
the
standard.
A
So
you
know,
sbml
is
an
xml
based
language
for
representations
of
species,
compartments
reactions
and
their
specific
properties.
So
this
is
a
chemical
jargon,
and
this
is
you
know
this
is
kind
of
one
of
these
things
that
they
use.
I
think
a
lot
for
metabolic
metabolic
networks,
so
you
know
they're
focused
largely
on
biochemistry
here
in
this
part
yeah.
So
sbml
is
initially
focused
on
biochemical
reaction
networks.
A
A
It's
hard
to
create
models
of
using
that
standard
for
regulatory
networks,
regulatory
networks.
People
generally
use
logical
models,
which
means
that
when
there's
some
criterion,
that's
met,
a
gene
is
expressed
or
it's
expressed
to
a
certain
level.
That's
that's
typically
the
standard,
so
people
are
developing
standards
for
that,
but
that's
a
different
standard
cell
ml
is
another
one
of
these
xml
based
standards.
A
This
is
a
more
generic
form
of
an
xml
standard
and
it's
often
used
in
other
to
model
other
types
of
networks,
creating
these
graphical
models
with
a
common
notation,
and
so
they
go
over
a
number
of
different
networks.
Here,
different
approaches,
process,
algebras,
petri
nets.
They
show
some
formalisms
with
a
visual
representation
here.
A
So
this
is.
These
are
boolean
networks.
Here
these
are
bayesian
networks.
These
are
petri
nets,
d
or
agent
based
models.
G
are,
let
me
see
g
or
cellular
automata.
Of
course,
f
are
rule
based
models
and
then
e
is
they're.
Just.
C
C
A
With
this
paper,
I
think
it's
you
know
if
you
want
to
know
more,
you
can
read
about
it,
but
it's
a
nice
review
of
that
area
and
I
think
it's
useful
here,
because
we
talk
a
lot
about
these
different
things
and
at
some
point
we
want
to
merge
them
together
and
bring
everything
together
in
a
way
that
we
can
interpret
for
the
whole
organism.
A
It's
it's
it's
to
note
that,
like
in
the
open
worm
foundation,
you
know
we
have
a
lot
of
parallel
projects
focused
on
like
movement
and
focused
on.
We
don't
really
focus
on
gene
expression
or
metabolism
very
much,
but
we
do
a
lot
of
stuff
with
you
know
bioelectricity
and
things
like
that.
So
you
know
we
don't
really
have
a
a
method
to
unify
everything.
We
kind
of
have
these
parallel
approaches
and
they're
kind
of
they
kind
of
fit
together.
A
A
A
One
of
the
people
was
amy
scheier,
who
did
some
work
on
looking
at
morphogenesis
as
this
process
of
pushing
and
pulling,
and
they
compared
it
directly
to
the
reaction
diffusion
model
of
turing,
so
the
reaction
to
f
diffusion
model
of
turing.
If
we
go
back
to
this
poster,
you
know
this
kind
of
pattern:
formation,
where
you
have
basically
something
that's
random
and
then
it
forms
like
dots
or
stripes,
or
something
like
that.
That
is
turning
reaction
diffusion.
A
This
is
a
process
where
chemical
morphogens
interact
and
form
these
gradients,
and
then
they
form
these
clumps
okay,
and
this
is
a
chemical
process
where
the
chemical
different
species
of
chemical
or
different.
A
You
know
morphogen
types,
kind
of
interact
and
form
a
boundary
between
one
another,
and
so
this
is
why
you
get
the
striping
or
these.
This
spotted
pattern
in
in
the
new.
This
alternate
model
that's
been
proposed.
That's
actually
not
what
happens?
A
What
you
have
is
you
have
this,
these
forces
that
occur
in
a
sheet
of
cells
and
they
force
like
these
buckles
in
the
surface,
and
then
those
buckles
are
reinforced
by
some
sort
of
activity,
so
it
could
be
like
developmental
genes
that
are
expressed
in
these
certain
areas
where
the
buckles
occur,
but
not
other
areas.
A
The
sheet
buckles
and
buckling
is
not
a
uniform
process
that
forms
these.
You
know
different
hills
and
valleys
and
then
those
areas
that
are
uplifted
as
it
were.
They
start
to
get
reinforced
by
gene
expression,
developmental
gene
expression
of
different
factors,
and
then
you
end
up
with
these
spots
or
these
ridges
or
whatever.
It
is
the
pattern
that
you
observe,
so
it's
a
very
different
way
of
viewing
morphogenesis,
and
so
these
are
some
papers
that
kind
of
go
in
the
same
direction.
A
So
this
is
another
paper
that
they
did.
I,
I
don't
think
I
talked
about
this
paper,
but
this
is
from
this
group
and
their
paper
is
called
emergent.
Cellular
self-organization
and
mechanosensation
initiate
follicle
pattern
in
the
avian
skin,
so
they
were
actually
looking
at
the
avian
model.
They
were
looking
at
skin.
They
were
actually
looking
at
dermal
formations,
so
they
were
looking
at
where
the
follicles
will
be
for
feathers.
A
So
the
follicles
are
points
on
the
skin
where
the
feathers
will
grow
out,
but
the
follicles
need
to
form
and
they're
different
from
the
rest
of
the
skin.
So
you
had
this
buckling
process
and
you
ended
up
with
you
know,
going
from
like
sort
of
a
homogeneous
field
to
this
inhomogeneous
field,
where
you
have
these
spots
that
are
going
to
become
something
in
this
case
follicles.
A
So
the
abstract
of
this
paper
reads
the
spacing
of
hair
and
mammals
and
feathers
and
birds
is
one
of
the
most
apparent
morphological
features
of
the
skin.
This
pattern
arises
in
uniform
fields
of
progenitor
cells,
and
these
are
these
epithelial
cells.
Epithelial
progenitor
cells
diversify
their
molecular
fate,
while
adapting
higher
order
structure
so
diversifying
their
molecular
fate
means
that
certain
things
are
expressed
in
some
cells
that
experiences
buckling
force
versus.
A
A
The
key
initiators
of
heterogeneity
are
dermal
progenitors,
which
spontaneously
aggregate
the
contractility
driven
cellular
pulling.
So
this
is
this
buckling,
I'm
talking
about
it's.
This
contractile
force
it's
pulling
at
the
cell,
the
sheet
of
cells,
and
it's
forming
these.
These
places,
where
you
can
get
these
these
these
these
spots.
A
Concurrently,
this
normal
cell
aggregation
triggers
the
mechanosensitive
activation
of
beta-catenin
and
adjacent
epidermal
cells,
initiating
the
follicle
gene
expression
program.
So
we
have
this
force.
That's
initiating
this,
this
gene
expression
cascade.
It's
like
one
factor,
that's
really
driving
us,
but
it's
driving
this
per.
It
turns
this
program
on
basically
taken
together.
This
mechanism
provides
a
means
of
integrating
mechanical
and
molecular
perspectives
of
organ
formation,
so.
C
A
A
You
can
see
these,
and
then
you
have
bmp2,
which
is
something
that's
being
expressed.
As
an
example
of
you
know
how
this
program
is
turned
on
in
these
locations,
so
bmp2
is
being
turned
on
in
this
image.
You
have
the
dappy
stain,
which
is
this.
Well,
you
have
that
b
and
e
kit
here
and
actually
you
know
you
have
beta
cayton
in
here
so
beta-cadent
and
dappy
together
in
this
image,
and
then
this
is
the
dappy,
some
laminar
marker
and
the
need
kid
hearing.
A
So
this
shows
the
different
things
that
are
being
expressed
in
this
process
over,
like
from
day
six
to
day
eight,
it's
a
very
tight
window
where
this
happens
in
this
day,
six
of
days
of
check
development,
which
I
don't
know
what
the
point.
C
A
It's
probably,
you
know
been
been
development
or
something
like
that.
It's
it's
not
it's
not
very
far
along
so
yeah.
This
is
another
example
here,
where
you
have
a
control
versus
something
that's
been
treated
the
treatment
actually
so
yeah
this
this
xav939.
A
This
actually,
I
think,
inhibits
the
formation
of
these
sharp
boundaries.
So
it's
like
something
they
can
do.
They
can
do
like
in
a
culture,
they
have
a
control
and
then
they
have
this
and
it's
inhibiting
the
formation
of
these
sharp
boundaries-
and
you
can
see
here
that
the
this
is
shown
in
the
stain
where
this
is.
This
is
the
result
of
inhibiting
boundary
formation
so
that
you
know
they're
able
to
do
this
for
just
within
the
chick
embryo
they're
able
to
show
you
know
they
in
these
kind
of
papers.
A
They
show
all
these
different
possibilities.
You
know
they
like
to
stain
the
sample
quite
a
bit
and
have
these
different
things
that
are
expressed
and
show
images
of
it,
because
it's
really
kind
of
you
know.
They'll
have
quantitations,
but
they'll
always
show
the
images
so
yeah.
So
it
goes
through
that
and
then
it
kind
of
goes
through.
The
puts
it
on
different
substrates
and
I
think
this
is
in
a
culture,
a
stiff
substrate
to
flexible
substrates.
So,
as
you
go
along,
the
substrate
makes
a
difference.
A
So
this
is,
they
want
to
show
kind
of
the
context
where
you
get
this
sort
of
pattern
formation,
and
so,
if
you're,
putting
these
samples
of
the
tissue
on
gel,
you
take
them
out
of
the
embryo,
and
you
put
them
on
a
gel
and
the
gel
is
of
different
sort
of
proper
properties
of
the
gel.
So
it's
stiff
versus
flexible.
A
It
allows
the
cells
low
traction
versus
high
traction.
You
get
these
differences,
so
the
substrate
or
where
the
sort
of
the
surface
on
which
this
occurs
matters,
and
so
there's
a
surface
of
the
embryo.
The
skin
is
just
one
part
of
it:
it's
it's
forming
on
other
parts
of
the
embryo,
so
there
there
are
forces
at
work
here
that
are
you
know
this.
This
process
requires
a
specific
physical
context
and
then
this
is
the
movement
of
beta
katana
into
the
nucleus
and
the
forming
primordium.
A
This
is
mechanically
triggered
in
upstream
of
the
primordium
gene
expression
program.
So
this
is
where
you
get
these
different
like
this
is
one
example:
you
have
this
gradient
of
traction
of
this
gradient
of
resistance,
and
if
you
go
to
different
points
on
this
continua,
you
get
these
different
effects.
A
So
you
get
clustering
or
you
get
dissociation,
and
so
you
can
see
that
the
dermis
compresses
epidermis,
so
the
dermis
is,
I
think,
on
top
the
epidermis
is
here
and
it's
compressing
the
epidermis,
this
force
of
the
dermis
compressing
the
epidermis
causes
translocation
of
beta
cadent,
and
so
you
can
see
this
in
the
cross
section
that
this,
as
this
epidermis
actually
is
on
top
dermis,
is
on
bottom
epidermis
starts
to
buckle,
it
compresses
the
dermis
and
then
the
dermis
has
there's
some
in
this
creates
forces
within
the
dermis,
and
you
start
to
get
this
pattern
formation,
so
pattern
formation
accelerates
as
these
forces
are
introduced.
A
So
that's
a
very
interesting
approach.
There's
some
other
papers,
though,
that
are
related
to
this.
This
is
another
paper
on
growth
and
form
of
the
gut,
and
this
kind
of
goes
through
some
aspects
of
dot
formation.
That
in
the
chick
embryo,
this
is
another
example
of
physical
forces
playing
a
role
in
development.
A
A
simple
physical
mimic
using
a
differentially
streamed
composite
of
a
pliable
rubber
tube
and
a
soft
latex
sheet
is
consistent
with
this
mechanism
and
produces
similar
patterns.
So
you
can
actually
produce
this
sort
of
morphogenetic
pattern
using
physical
analogs,
which
is
interesting,
and
so
they
built
a
mathematical
model.
The
predictions
of
their
theory
are
quantitatively
consistent
with
observations
of
intestinal
loops
at
different
stages
in
the
development
of
the
czech
embryo.
A
A
A
You
get
these
things
that
that
form
layers,
one
on
top
of
the
other.
There's
some
stretching
and
relaxing,
and
you
get
this
patterning,
this
looping
formation
and
it's
affecting
the
tissue
underneath
it
as
well
so
to
construct
a
rubber
model
of
looping.
A
thin
rubber
sheet
is
stretched
neoformally
across
its
length
and
then
stitched
to
a
straight
unstretched
rubber,
tube
along
its
boundary,
the
differential
strain
mimics,
the
differential
growth
of
the
two
tissues,
so
these
tissues
are
growing
at
different
rates.
A
This
is
more
morphometric
and
mechanical
measurements
of
check
cut.
So
this
is
again
this
looping
they're
able
to
do
some
quantification
of
the
model
and
make
some
predictions,
and
then
this
is
more
of
this
modeling,
where
they
can
make
predictions
for
loop,
shape,
size
and
number
of
three
stages
in
gut
development,
so
they're
able
to
do
this
for
different
stages
of
development
and
then,
finally,
this
is
an
example
of
chick
at
e12
quail
at
e12
and
e12
is
just
the
developmental
day,
finch
and
mouse.
A
So
you
can
see
this
in
different
species
where
you
can
make
comparative
predictions
for
looping
parameters,
so
this
actually
happens
in
a
wide
range
of
guts
and
they're
able
to
predict
for
each
of
these.
They
use
the
chick
model
to
develop
this
technique
or
this
model,
and
then
they
apply
to
other
species.
A
So
that's
a
nice,
like
example
of
this,
in
sort
of
a
mathematic,
more
mathematical
context.
This
paper,
of
course,
also
talks
about
vy.
It's
called
velification,
and
so
this
is
by
a
group
of
people,
including
amy
shire,
some
other
people
in
here,
and
they
kind
of
go
in
the
same
direction.
So
the
abstract
of
this
paper
reads:
the
vui
of
the
human
and
chikka
are
formed
in
similar,
stepwise
progressions,
where
in
the
mesenchyma
and
attached
epithelium
first
fold
into
longitudinal
ridges
than.
A
Pattern
then,
finally,
individual
voi,
we
find
that
these
steps
of
velification
depend
upon
the
sequential
differentiation
of
distinct,
smooth
muscle
layers
of
the
gut.
So
you
see
the
same
process
as
the
looping.
You
have
these
different
layers,
they're
growing
at
different
rates,
they're
they're,
changing
their
shape
in
different
ways,
and
then
the
change
in
shape
in
one
layer
is
affecting
the
shape
in
other
layers,
which
so
this,
this
sort
of
the
sequential
differentiation
leads
to
a
restriction
of
the
expansion
of
the
growing
endoderm
and
mesenchyme,
which
are
the
other
layers
generating.
A
A
We
have
this
paper
on
self-assembly
of
tessellated
tissue
sheets
by
expansion
and
collision.
This
is
a
different
group
of
people,
and
this
abstract
reads:
tissues
do
not
exist
in
isolation;
they
interact
with
other
tissues
within
and
across
organs,
while
cell
cell
interactions
have
been
intensely
investigated.
A
Less
is
known
about
tissue
tissue
interactions,
so
cell
cell
interactions
are
when
cells
are
behaving
interacting
with
one
another.
We
also
have
tissue
tissue
interactions,
which
are,
of
course,
very
important.
As
we've
just
mentioned
here,
we
studied
collisions
between
monolayer
tissues,
which
are
things
that
form
a
single
layer
with
different
geometries
sultan
cities
and
cell
types.
So
now
they're
not
interested
in
the
interactions
between
layers
per
se,
they're
interested
in
monolayer
tissues
that
have
different
shapes
and
different
densities
of
cells
and
types
of
cells,
which
can
actually
act
very
much
like
this
multi-layered
structure.
A
First,
we
determine
rules
for
tissue
shape,
changes
during
binary
collisions,
so
they
take
cells
one
at
a
time
they
collide
them.
They
describe
complex
cell
migration
at
tri-tissue
boundaries.
So
if
you
have
different
tissues
that
are
coming
together,
you
know
you
can
describe
the
sort
of
how
cells
migrate
in
that
boundary
space.
A
Next,
we
propose
that
genetically
identical
tissues
displace
each
other
based
on
pressure
gradients.
So
these
should
be
all
identical
tissues.
They
shouldn't
have
any
you
know:
genetic
mutations
between
them
displace
each
other
based
on
pressure
gradients,
which
are
directly
linked
to
gradients
and
cell
density.
So
this
is
again
we're
dealing
with
gradients,
we're
not
dealing
with
chemical
gradients,
we're
dealing
with
physical
gradients.
A
A
Finally,
we
introduced
tysolate,
which
is
the
design
tools
for
self-assembling
complex
tessellations,
from
arrays
of
many
tissues.
We
use
cell
sheet
engineering
techniques
to
transfer
these
composite
tissues
like
cellular
films,
so
tessellations
are,
are
like
different
tilings
of
of
usually
in
like
mathematics.
They
use
the
term
tessellation
to
talk
about
how
you
can
pack
different
shapes
into
a
space,
a
continuous
space,
and
you
know
there
are
different
ways.
A
When
you
get
two
circles
that
collide
or
two
like
blobs,
they
form
this
like
it
looks
like
an
ice
cream
cone
almost
and
the
same
thing.
With
these
blobs
of
different
sizes,
it
almost
looks
like
a
shoe
print
or
a
light
bulb,
and
then,
if
you
have
a
long
rod
and
a
blob,
they
come
together
and
they
form
this
shape,
and
it
actually
looks
like
something.
That's
from
the
embryo.
I
mean
you
know
they're
all
from,
I
guess
an
embryo,
but
they
have
these
distinct
shapes
and
they're
predictable.
A
That's
what
they're
saying
here
so
you
know
I'm
not
going
to
get
too
much
into
this
paper,
I'm
just
going
to
point
out.
They
did
a
lot
of
like
engineering
type
analysis,
and
it's
really,
you
know
interesting
work.
It
kind
of
follows
up
very
nicely
with
some
of
the
things
we've
talked
about,
so
these
are
tessellations
and
again
these
are
shapes
that
are
like
packed
into
a
surface.
So
this
is
a
this
type
of
thing
here
is
a
a
target.
A
Tri-Color
tessellation-
and
this
is
just
says,
like
three
different
colors
and
three
different
sort
of
sides-
those
like
these
cubes
that
are
packed
into
the
space.
This
is
another
example
of
a
tessellation
where
you
have
a
repeating
pattern,
but
they're
packed
in
so
it's
continuous.
There
are
gaps
in
between
them.
This
is
another
example
of
a
tessellation
where
you
combine
these
blobs
into
a
continuous
form
and
so
on
and
so
forth.
A
So
that's
you
know.
That's!
I
wanted
to
follow
up
on
that
topic
and
I
think
I
did
a
pretty.
I
think
that
was
a
pretty
good
representative
sample
of
what's
going
on
with
a
lot
of
these.
A
A
lot
of
this
work,
so
we
have
one
comment
here.
This
is
a.
C
A
B
Just
that
the
bulk
modulus
is
mostly
due
to
water
in
cells.
It's
it
dominates
over
the
elasticity
and
viscosity.
A
A
Experiments,
I
think
you
know
some
of
the
things
that
you
measure
in
culture-
maybe
don't
map
to
the
organism
if
you
actually
do
things
in
vivo,
which
is
to
do
it
in
the
organism
and
observe
the
thing
in
the
organism.
A
C
Just
to
add
to
that
that
you
know
that
water
can
can
also
take
different
forms
when
it
interacts
with
biological
structures,
and
you
know
I'm
just
thinking
class
rates
and-
and
I
think
there's
some
other
papers
that
suggest
that
you
know
that
the
water
should
be
modeled
or
we
should.
C
We
should
be
careful
in
our
assumptions
about
movement
inside
the
cell.
I
mean
I'd
love
to
hear
someone
who's
got
more
more.
You
know
knowledge
on
this
speak,
but
but
I
think
it's
really
interesting
that
water
can
actually,
you
know,
behave
differently
and
we're
not
just
talking
about
freezing
yeah
yeah.
A
A
So
they're
they're
doing
this
very
focused
study,
but
sometimes
there
are
a
lot
of
things
that
are
affecting
and
I
talked
earlier
about
the
systems
biology
models.
You
know
that
that
sort
of
sort
of
thing
where
you
not
only
have
these
different
sort
of
processes
like
gene,
expression
or
proteomics,
but
you
also
have
these
things
like
physics
and
water
and
things
that
are
pretty
standard
in
biology,
but
they
don't
have
a
model
for
it
and
they
don't
fit
into
these
things.
B
C
C
B
C
I
I
also
would
like
to
know
more
about
pink
petri
nets.
C
A
Yeah,
it's
like
petri,
nuts
and
reservoir
networks.
Those
are
two
things
that
are
really
kind
of
like.
I
can't
really
wrap
my
head
around
fully,
but.
C
A
Okay,
so
thanks
for
attending
next
week,
we'll
have
more
updates
on
gsoc
and
more
papers
and
more
discussion
so
have
a
good
week.