►
From YouTube: DevoWorm (2021, Meeting 3): Issues board, evolution and developmental theory, diffusion on networks
Description
DevoWorm (2021, Meeting 3): Issues board, Member Achievements, Submission to Evolution conference, Agent-based simulations of network diffusion, deep evolutionary history figure and the Boring Billion, papers on ML/Development and the History of Developmental Theory. Attendees: Susan Crawford-Young, Bradly Alicea, Krishna Katyal, Jesse Parent, and Shruti Raj Vansh Singh
A
A
B
B
A
There
we
go
jesse.
How
are.
A
A
A
C
A
A
Well,
I
think
you
can
probably
use
the
chat
for
just
you
know
I'll
check
I'll
keep,
checking
the
chat
we'll
go
from
there.
A
Sound
card,
I
think,
was
having
problems
in
previous
meetings,
so
I
don't
know
why
it's
not
working.
A
Okay,
welcome
to
the
meeting
and
we're
going
to.
I
changed
the
meeting
venue
because
I
already
used
this
for
another
set
of
meetings.
I
like
fujitsi
at
some
point.
I'm
gonna
try
to
upgrade
it
to
something
where
I
can
control
the
interface
better,
like
jitsi
is
an
open
source
meeting
space
and
it
works
generally.
Well,
the
if
you
want
to
share
your
screen,
there's
a
if
you
want
to
do
the
chat,
there's
a
button
at
the
lower
left
of
the
screen.
It's
like
a
thought
bubble.
A
A
I
think
it
was
just
hitting
the
hand
yeah.
It
was
okay,
it's
a
hand
raising,
so
that
works.
So
you
can
see
that
in
the
in
the
corner
there-
and
I
guess,
if
you
set
this
up
on
a
server-
you
can
install
custom
elements
so
that
might
come
in
the
future,
but
I
for
right
now
we'll
just
use
this
setup.
A
So
a
lot
of
things
to
get
to
today
we
have
some
things
on.
I
have
some
interesting
papers.
I'd
like
to
make
some
congratulations
to
mayuk.
I
don't
know.
I
was
kind
of
hoping
that
krishna
would
come
to
the
meeting
and
he
was
gonna
present,
but
he
actually
has
some
slides
that
he
sent
me
on
something
he
wants
to
submit
to
the
evolution
conference.
A
We'll
look
at
those
and
then
maybe
we'll
go
over
the
task
board,
and
so
that's
that's
what
I
have
for
today.
First
of
all,
anyone
want
to
bring
up
anything
that
they
have,
that
they
from
last
meeting
or
from
previous
meetings
that
they
wanted
to
revisit.
A
A
This
is
pretty
positive.
There
are
some
small
edits
to
be
made
as
required
by
the
reviewers,
so
we
have
some
things
like
you
know.
Well,
there's
probably
some
of
the
things
that
you
hit
on
in
the
edits,
but
also
some
things
they
wanted
to
change.
For
you
know
I
assume
acceptance
and
publication.
So
that's
good,
we'll
we'll
be
revisiting
those
soon.
A
A
Hello-
and
you
have
your
camera
on
today,.
A
Nice
to
see
you
so
yeah,
so
we
were
just
talking
about
some
of
the
like
we're
talking
about
the
periodicity
paper,
so
it's
it's
in
pretty
good
shape
it
it
it's
probably
going
to
be
accepted.
We
have
to
do
some
editing
on
it
and
so
we'll
be
talking
about
that
in
another
meeting
coming
up
so
yeah.
A
Is
there
anything
you?
You
had
any
announcements
that
you
had
from
like
past
meetings
or.
A
Okay,
so,
okay,
so
then
let's
move
on
so
we
haven't
done
this
this
year.
Yet
I'll
share
my
screen
and
we'll
do
the
major
tasks
for
2020,
which
is
our
task
board.
Can
you
see
my
screen.
B
A
Okay,
so
this
has
been
a
while,
since
we've
revisited
this
board,
we
have
a
bunch
of
things
and
to
do
we
have
get
data
for
citizen
spherical,
image
data-
I
guess
that
was-
or
that's
probably
not
a
good
way
to
phrase
that,
but
this
is
something
to
do
with
the
axolotl
data,
and
I
guess
susan
said
that
there
was
more
coming
and
we
have
some
that
I
have
in
my
possession.
A
So
is
there
an
update
on
that
susan?
Oh
check
the
chat.
A
I
think
that
was
oh
november
30.,
so
I
mean
you
know
we're
when
we
actually
did
come
up
with
a
summer
of
code
project
for
the
axolotl
data,
so
we
might
be
able
to
get
some
people
interested
in
that
and
move
along
on
that,
but
we
already
have
data
for
it.
You
know
to
start
off
with,
so
it's
not
like.
We
need
to
do
it
right
away,
but
not
like.
We
need
to
have
it
by
then.
Oh
there's
krishna,
hello,
krishna,
you're,
muted,
by
the
way.
A
A
I
haven't
really
explored
that,
yet
we
can,
if
you
wanna,
if
someone
wants
to
think
more
about
that
you're
welcome
to
do
so,
but
that
might
be
a
good
way
to
sort
of
get
pulled
together,
a
lot
of
the
machine
learning
stuff
that
we've
been
doing
and
some
of
the
other
stuff.
A
I
don't
know
what
goes
into
building
a
kindle
book.
Maybe
what
just
having
like
a
electronic
book
that
people
can
access,
so
they
can
access
everything
in
one
place,
but
that
I
mean
I
don't
know
how
much
work
that'll
take,
but
you
know
we'll
probably
get
to
that
in
the
next
couple
weeks
or
so
network
science
submission.
This
is
something
there's
a
conference
called
complement.
That
is
maybe
a
target
conference.
As
I
mentioned
last
week
we
had
we.
A
I
showed
you
the
the
document
of
deadlines,
which
I
think
is
in
the
slack
channel.
So
if
you
want
to
look
that
over
one
of
those
submissions
places
as
complement,
which
is
a
complex
networks
conference
and
we
can
submit
something
there,
if
we
have
ideas
about
networks
in
that
area,
we
can
we
can
submit
there
and
then
this
annotated
bibliography.
So
we've
been
started
talking
about
that
over
the
past
several
months.
A
This
is
still
something
up
in
the
air
if
people
are
interested
in
taking
that
on,
let
me
know
we've
been
doing
we're
thinking
of
doing
this
in
my
other
group
for
some
other
for
a
topic
called
developmental
ai.
So
this
might
be.
You
know
something
where
we'll
have
crossover
with
that.
A
So
in
hold,
we
have
updates
on
axolotl
data
and
analysis.
I
think
susan
kind
of
told
us
what
that
was
recruit
people
as
diva
learned
contributors.
That's
you
know
again,
that's
something
that
we
can
do
on
an
ongoing
thing,
but
we
don't
have
any
events
that
will
meet
that
goal.
A
A
A
So
this
is
something
that
we
can
do.
If
people
are
interested
in
making
a
video
tutorial,
you
can
go
ahead
and
inquire,
maybe
even
make
one,
and
then
we
could
review
it
and
put
it
up
on
the
youtube
channel.
A
So
a
lot
of
these
things
too,
the
embryo,
visualization
cree
embryo
model
and
blender
that's
sort
of
a
very
long
standing
thing
that
we've
been
wanting
to
do,
but
for
to
remind
people.
We
wanted
to
build
a
virtual
embryo
and
blender,
which
is
a
3d
modeling
software
and
we'd,
eventually
like
to
build
something
for
the
open
worm
docker,
which
is
a
software
container,
that
you
can
download
and
experience
all
the
open
worm
simulations
in
one
place,
and
we
don't
have
a
developmental
version
of
that.
A
So
if
we
could
have
something
that's
developmental
and
scope,
that
would
be
great.
So
I
think
all
these
are
just
kind
of.
I
don't
have
any
updates
on
these
update,
diva
or
ml
lectures.
A
This
is,
of
course,
something
that
I
was
doing
towards
the
end
of
last
year,
but
we
do
have
this
diva
warm
ml
course
from
2019
and
that
content
is
in
need
of
being
updated.
So
if
people
have
ideas
about
updating
that,
we
can
put.
A
Issue
73.
this
work
on
talk,
slides
for
osf,
divoler
and
flash
talk.
That's
something
I'm
going
to
get
get
to
in
a
minute
devon
bibliography
that
kind
of
goes
with
this
annotated,
bibliography
thing,
that's
something
that
we
can
update
as
we
review
papers
and
things
like
that,
the
bacilluary
psychophysics,
that's
something
that
we
that
we
we
owe
someone
a
paper
on
this
in
april.
So
we
need
to
get
back
to
this,
but
that's
sort
of
on
hold
for
now,
but
I
mean
put
it
in
hold.
A
I
guess
too,
let's
see
axolotl
embryo
animations
and
segmentation,
that's
something
that
and
it
goes
along
with
the
axolotl
data.
So
once
we
get
that
figured
out,
all
these
issues
can
be
addressed.
A
The
gsoc
2021
proposal,
I
think,
that's
finished,
because
we
submitted
those
to
malen.
So
now
we
wait
and
get
people
interested
in
making
proposals.
For
that
recap,
presentation
for
2020,
that's
finished
great
evil
and
paper
from
josh
submission.
A
That's
a
yeah.
That
would
be
an
action
item
because
it
needs
to
be
done,
but
you
know
it's
kind
of
on
the
back
burner.
Still,
I
don't
know,
maybe
hold
for
that
and
then
in
progress
we
have
this
divorce
bibliography
that
will
grungy
embryo
readings.
I'm
not
sure.
Oh.
This
is
the
thing
that
it
was
a
long
time
ago
where
we
talked
about
a
lagrangian
embryo.
That's
sort
of
on
hold,
create
a
theory
layer
for
deviler
and
that's
an
on-hold
thing.
A
Well,
there's
a
periodicity
in
the
embryo
paper.
So
that
is
actually,
I
would
say,
that's
an
action
item,
because
the
paper
is
pretty
much.
You
know
it's
done
and
it's
been
pre-printed
and
we
have
this
first
set
of
reviews,
but
we
just
have
to
now
have
to
address
the
review.
So
maybe
we
can
add
a
card
here
that
says.
A
All
right
and
then
that'll
be
in
progress,
and
then
this
issue
when,
when
this
is
done,
those
two
issues
will
be
addressed.
Then
we
have
a
diva
worm,
bibliography
endnote.
I
think
that
goes
along
with
this
other
set
of
issues
with
the
annotated
bibliography
and
the
bibliography.
A
C
A
So
we
have
complexity
measures,
which
is
an
ongoing
thing.
That's
where
we,
you
know,
talk
about
different
measures
of
complexity
and
development
and
that's
sort
of
an
open-ended
issue,
and
then
this
bacilluary
non-normal
cognition
paper
is
the
same
thing
pretty
much
as
this
one
here.
A
So
maybe
we
should
put
follow-up
on
bachelorette
psychophysics
is
in
progress,
and
that's
that's
basically
addressing
that
paper.
This
issue
is
more
about
like
kind
of
following
up
on
the
measures,
but
I
think
we
actually
have
that
done
by
the
way,
because
it's
something
that
was
put
into
the
talking
that
went
to
neuromatch.
A
So
then
we
have
these
things
that
are
off
the
radar.
So
these
are
things
that
we've
had
on
the
board
that
I've
been
there
for
a
long
time,
and
I
ca-
I
don't
know
where
to
put
them.
So
these
are
things
like
there
was
a
a
thread
on
computed
tomography
via
cloud
computing.
A
That's
in
one
probably.
A
Now
that
was
brought
up
and
it
never
got
followed
up
on
no
one
ever
took
it
took
us
up
on
the
offer
to
do
anything,
so
it
that's
off
the
radar.
But
if
people
want
to
address
that
issue,
it's
open
krishna
was
going
to
review
a
paper.
I
can't
remember
what
that
was
about,
but
that's
something
that
krishna.
If
you
have
want
to
follow
up
on
that,
you
can
put
a
note
in
this
issue
44.
A
this
game
theory
for
developmental
processes.
This
is
a
one
of
the
open
papers
that
we
have
in
the
open
papers
list
it's
kind
of
been
frozen
in
time
for
a
while,
because
it
kind
of
been
doing
a
lot
of
other
things.
A
But
if
that's
that's
an
an
interesting
idea
that
I
mean
there's
an
outline
that
exists
in
that
document
right
now
and
it
needs
to
be
fleshed
out
more
and
it
might
actually
have
some
tie-ins
with
some
of
the
things
we're
doing
in
my
other
group
regarding
developmental
ai.
So
jesse
knows
what
I'm
talking
about
on
that,
and
you
know
that's
something
that
we
can
revisit.
A
Maybe
in
the
next
couple
weeks,
there's
actually
a
lecture
in
the
evil
worm
ml
materials
called
it's
it's
part
of
the
gans
lecture
or
in
that
week,
where
we
talk
about
game
theory
because,
as
it
turns
out,
gans
are
sort
of
inspired
by
game
theory
and
there's
a
there's,
a
interesting
story
there.
But
in
any
case
this
might
be
an
interesting
thing
to
revisit,
follow
up
on
neural
neural
organoids
number
30..
A
I
just
talked
to
someone
who
I
would
talk
to
last
year
about
a
collaboration
that
involve
do
building
computational
models
of
systems.
They
call
neural,
organoids
and
neural.
Organoids.
Are
these
structures
where
you
get
a
bunch
of
stem
cells
and
you
have
them
in
like
a
3d
cell
culture,
environment
and
you
grow
them,
you
you
send.
You
know
you
get
them
to
differentiate
into
neurons,
and
then
they
grow
these
they're
sort
of
like
embryoid
bodies,
so
they
have
a
certain
shape.
A
You
know
they're
they're
not
really
differentiated
into
a
formal
organism
but
they're
sort
of
differentiated
tissues
in
a
blob
and
so
they're
using
these
as
model
systems
for
studying
a
number
of
things
that
neural
systems
do,
and
so
I've
been
talking
with
a
couple
people
about
this
over
the
last
year
and
a
half
maybe-
and
it's
been
pretty
low
key.
I
haven't
really
involved
anyone
else,
but
maybe
the
answer
is
just
to
open
this
up.
A
So
we'll
be
maybe
I'll
talk
about
this
a
little
bit
more
in
the
in
future
meetings.
Basically,
the
idea
is
to
sort
of
get
our
hands
around
a
computational
model
of
this
of
neural
organoids,
and
then
you
know
maybe
collaborate
with
some
people
who
collect
data
in
in
vivo
systems
where
they're
you
know
able
to
do
experiments.
A
So
that's
something
that
I
think
it
would
be
exciting
if
we
keep
it
off
the
ground,
but
it's
just
been
off
the
radar
this
paper-
oh,
this
is
something
jesse
wanted
to
review
at
one
time.
That
was
something
that
was
scale-free
biology.
That
was
a
long
time
ago
now,
but
if
anyone's
inter,
if
he's
interested
in
that
still,
we
can
do
that.
There's
also
this
compositionality
paper,
which
is
something
that
I
don't
think
it's
even
in
the
open
papers.
But
it's
something
that
I
was
thinking
about
for
developmental
biology.
A
A
So
so
we've
got
okay.
So
now,
let's
move
on
from
that,
I
think
if
there
are
no
well
okay,
we
have
some
things
in
the
chat.
First
of
all,
so
jesse
says
I'd
be
interested
in
the
kindle
book.
Okay,
yeah,
that's
good!
A
I'm
gonna
try
to
get
some.
I'm
gonna
try
to
maybe
move
on
that
a
little
bit
or
maybe
put
something
in
the
slack
on
this,
where
we
can
organize
our
thoughts
about
how
to
put
this
together,
I
don't
really
have
experience
with
kindle
books.
So
if
someone
knows
how
to
do
that
or
if
you
could,
you
know,
we
have
to
think
of
like
a
format
to
put
things
together
and
but
we
can
do
that
and
then
susan's
now
we
have
two
s's.
Yes,
we
have
surety
and
susan.
A
So
the
you
know
the
jet
sea
interface,
when
you
have
your
camera
off
as
a
little
letter
icon
for
your
name,
so
jesse
says
already
annotated,
bibliography
I'll,
be
interested
in
both
parts
of
those
are
in
working
parts
of
it.
Yeah
so,
like
I
said,
there's
a
lot
of
overlap
between
this,
like
what
we
do
in
divorm
and
this
initiative
developmental
ai
in
my
other
group
and
krishna,
probably
be
interested
in
this
too,
where
you
know
we're
talking
about
building
a
sort
of
an
artificial
intelligence
system
based
on
developmental
principles.
A
So
there's
going
to
be
there'll,
be
citations,
that'll
bridge
the
two
groups
so
we'll
see
about
that.
I'm
interested
in
measuring
them
with
oct
so
interested
in,
measuring
what
I
guess
it
was
something
organoids:
okay,
yeah
the
organoids
thing.
Yeah
it's
it'd
be
interesting.
A
You
know
there's
some
groups
doing
like
you,
have
to
have
an
experimental
set
up
and
I
don't
have
access
to
that
or
I
don't
know
if
susan
could
do
it
because
it's
you
know
it
takes
a
pretty
specialized
setup,
but
you
know
if
we
could
find
a
collaborator,
we
might
be
able
to
do
some
things,
and
I
know
they
do
a
lot
of
measurements
with
you
know
some
of
the
things
that
like
voltage,
dies
and
things
like
that
at
least
what
I've
seen
in
the
literature
so
there's
a
whole
literature
measuring
things
in
organoids
and
there's
room,
of
course,
for
improvement.
A
You
can
use
a
lot
of
microscopy
techniques
to
do
this
stuff.
So
that's
that's
something
we
can
talk
about
later.
Jesse
says
I
forgot
about
the
scale
free
paper.
I'd
like
to
do
at
some
point,
but
yes,
there
are
other
things
ahead
of
it.
For
me,
okay
and
then
susan
says
I
have
a
book
I
might
want
to
put
out
as
a
kindle
book
when
you
find
out
how
to
do
it.
Yeah.
So
we'll
follow
up
on
how
to
do
that.
A
I'm
sure
it's
not
difficult
to
I'm
just
making
myself
a
note
here
how
to
do
and
plan
so
plan
plan
of
attack.
You
know,
so
we
can
figure
out
what
what'll
be
in
it
and
that'll
go.
You
know,
go
hand
in
hand,
maybe
with
the
annotated
bibliography
or
whatever.
So
we'll
have
this.
You
know
at
the
end
of
the
year.
A
Maybe
we'll
have
this
nice
collection
of
things
that
we
can
put
out
as
a
book,
and
you
know
people
can
use
it
as
a
reference,
it's
kind
of
like
what
what
happened
with
divorm
ml.
You
know
that
was
something
where
kind
of
every
week
we
did.
I
I
did
a
lecture
almost
every
week
and
then
I
had
other
people
contribute
small
lectures
where
they
did
like
specialized
topics
at
where
you
know
people
brought
up
papers
to
the
meeting
and
we've
reviewed
them.
A
So
if
you
go
through
the
syllabus
you'll
see
that
I
don't
know,
maybe
that's
a
good
way
to
do
it.
You
know
have
like
a
chapter
where
you
do
delve
into
a
topic
a
little
bit
and
I
don't
know,
but
definitely
that's
one
model
to
do
it.
A
So
again
the
major
tasks
we
review.
If
we
want
to
add
any
issues
you
can
join
the
github
organization
or
join
the
github
repository,
propose
an
issue
or
answer
an
issue
there
on
github,
okay.
So
the
next
thing
I
want
to
talk
about
so
krishna,
actually
I'll
talk
about
your
slides
next,
so
you
sent
me
a
link
to
some
slides
on
evolution
or
the
thing
you
wanted
to
present
at
the
evolution
conference.
A
Okay,
yeah
there,
it
is
well
I'll,
bring
up
the
slides
and
I'll
go
through
them,
and
then
you
can
tell
me
if
the
if
you
know
your
comments
are
so.
C
A
We're
gonna
do
a
talk
here
or
propose
this
talk.
The
winner
takes
all
where
the
winner
gets
killed,
and
so
the
first
slide
is
an
ecosystem
designer.
A
A
C
Okay,
so
it's
regarding
the
evolution
of
conference
that
you
were
talking
about
in
the
previous
meeting,
so
it's
it
discusses
that.
Can
I.
A
Share
my
screen:
oh
yeah,
let
me
let
me
switch
off
here.
I'll
switch,
my
screen
off.
A
C
C
So
here
it
is
like,
for
example,
we
are
having
two
species
that
are
competing
with
each
and
every
within
with
each
other,
and
we
say
that
due
to
some
favorable
circumstances,
a
specie
a
gets,
you
know,
let's
get
the
edge
over
the
species
so
for.
C
And
or
dogs
fighting
with
each
other,
and
you
do
let
it
say
that
the
temperature
is
favorable
to
the
dogs
or
whatever
means
the
reason
the
dogs
get
an
edge
over
the
cat,
so
they
started
to
you
know
their
their
population
starts
to
get
increasing
under
that
population
starts
to
decreasing.
So
here,
if
we,
you
know
in
terms
of
darwin
philosophy,
survival
is
the
form
that
will
lead
the
most
copies
of
itself
in
successive
generation.
So
where
the
cats
are,
you
know,
are
getting
marginalized
and
the
dog
population
increases.
C
C
This
fight,
but
the
winner
gets
killed,
so
it
it
was
found
in
an
eukaryotic.
It
was
a
study
that
happened
with
diatoms,
so
the
when
they
get
kills.
So
if
you
want
to
say
that
okay.
C
For
example,
the
dogs
are
flourishing,
so
shall
be
the
cats
because
they
are
praying
on
the
dogs
and,
if
their
praise
flourishing,
they
should
also
flourish.
So
it's
simple
that
if
they
both
are,
you
know
trying
to
prey
on
each
other,
and
it's
simple
that
if
one
perishes
the
other
perish,
so
there
is
an
equilibrium
between
the
two
of
them.
C
So
it
says
that
our
equation
is,
for
example,
if
like.
C
If
the
population
of
mice
increases
the
population
of
cats
also
increases
was
you
know,
cat
praise
on
mice,
but
if
the
population
of
mice
decreases
due
to
scarcity
of
food,
the
cat
population
will
also
decrease.
So
it's
a
relationship
between
the
prey
and
the
predator
so
but
in
practice,
models
sure
that
the
combination
of
competition
and
relation
eventually
made
all
the
species
square
extreme,
so
how
it
can
be
possible
that
this
theory
makes
a
perfect
sense
that
and
now
certainly
with
evidence.
C
We
see
that
all
the
species
that
followed
it
are
extreme.
So
there
is
a
new
hypothesis
by
nature
goldfield,
and
she
is
you
know
it
was
so.
What
does
it
states
that
we
have
two
species
that
are
competing
with
each
other
and
the
one
species
tries
to
win
the
population
of
the
species
increases,
so
does
the
population
of
the
so
so
does
the
population
of
the
other
species
that
that
you
know
it
mutates
in
such
a
way
that
how
can
I
put
this
in
towards
that?
C
It
starts
to
invade
its
predator.
So
if
the
prey
predators
are,
you
know
getting
bigger
in
population,
the
prey
will
also
somehow
evolve
that
it
would
be
able
to
evade
the
predator
and
so
will
be
the
predator
evolving
so
that
it
can,
you
know,
hunt
the
prey
in
a
better
manner.
So
this
is
a
never
ending
cycle
of
improvement
that
helps
every
you
know:
species
to
survive.
C
A
Yeah,
it
looks
pretty
good
yeah
thanks
for
explaining
so
yeah.
I
think
that
would
be
a
good
idea
and
I
don't
know
you
know
if
we
want
to
ex
well,
I
mean
there
are
ways
you
can
approach
this.
One
thing
is
just
to
have
like
sort
of
a
theoretical
approach.
A
We
might
also
consider
using
some
sort
of
digital
evolution
approach,
which
of
course,
we
haven't
talked
about
too
much
in
the
group,
but
involves
a
you
know
where
you
simulate
evolution
in
a
program,
and
you
can
set
up
your
problems
so
that
you
have
your
predator
and
your
prey,
and
then
you
can
simulate
this.
A
You
could
use
like
you
know
something
like
avita,
which
is
a
program
very
that
has
specific
parameters
for
like
mutation
and
and
reproduction
or
you
can
use
something
like
an
agent-based
model
where
you,
I
think
you
can
specify
the
same
parameters,
but
it's
not
as
powerful,
but
in
any
case
I
think
we
can
address
the
core
issue
there,
which
is
the
hypothesis
that
you
mentioned.
A
We
could
do
that
or
we
could
approach
it
from
a
theoretical
point
of
view,
so
just
kind
of
like
maybe
updating
the
theory
or
thinking
about
you
know
how
it
applies
to
different.
You
know:
scenarios
or
animal
species.
I
mean
there
are
a
lot
of
ways
we
can
approach
it,
so
why?
Don't
we
I'll
I'll
think
about
it?
Some
and
you
can
think
about
it.
Some,
and
so
I
mean
the
submission
for
that
would
be,
I
think,
either
march
or
april
1st.
I
can't
remember
what
it
says.
C
A
And
we
could
talk
about
that.
Yeah,
like
the
evolution
conference,
is
pretty
broad,
but
there
are
certain
sort
of
tracks
it
needs
to
fit
into
to
get
it
to.
A
C
A
A
You
don't
see
anything
in
the
chat,
but
if
you
have
something
you
can
put
it
in
the
chat,
then
we
can
address
it
later.
Well,
thank
you
krishna
for
that.
That
was
pretty
good,
and
so
that's
that
that's
something
that
we'll
be
kind
of
making
into
it.
A
We
almost
have
to
make
it
into
a
so,
but
we
should
also
I
like
this
idea
of
putting
it
in
the
slides
first
and
then
you
know
we
can
like
kind
of
bounce
back
and
forth
between
that
and
making
the
actual
abstract,
which
actually
helps
you
later,
because
then
you
don't
have
to
worry
too
much
about
making
the
slides
so,
okay,
so
very
good.
The
next
thing
I
wanted
to
address
here
is
major
tasks.
A
We've
done
that
I
want
to
go
to
this
figure
for
the
boring
billion,
which
is
so.
This
is
something
that
I'm
working
on
with
dick
gordon
and
george
george
mikolovsky.
Actually,
we
have
some
comments.
A
Okay,
so
this
is
okay,
so
this
is
the
the
slides
here
for
christmas
presentation,
krishna
and
myself,
it's
presentation
and
then
just
requested
access.
Okay,
so
let's
see
so
this.
This
is
the
just
want
to
share
my
screen
again,
okay,
so
this.
A
On
for
this,
so
this
is
a
paper
that
they're
or
we're
working
on,
and
I
was
tasked
with
making
a
phylogeny
of
early
life,
and
I
did
some
research
on
this
and
this
isn't
the
flashiest
figure
in
the
world.
A
This
is
for
10
to
the
ninth,
that's
billion,
and
so
we
go
from
that
and
we
can
and
we
look
at
the
first.
This
is
the
first
event
where
you
get
a
division
between
bacteria
and
non-bacteria,
and
so
this
creates
a
common
ancestor
for
bacteria,
a
common
ancestor
for
non-bacteria.
A
So
they
live
in
ocean
vents
that
are,
you
know,
hundreds
and
hundreds
of
degrees
centigrade,
and
they
they
can
survive
in
extreme
conditions,
whereas
bacteria
are
the
garden
variety
bacteria
that
exist
on
every
surface,
at
room
temperature
and
so
there's
a
difference
between
those
two
groups
and
they
split
here
at
this
at
2.5
billion
years
ago,
then,
at
2.4
billion
years
ago
there
was
something
called
the
great
oxygenation
event,
which
is
where
he
started
to
get
oxygen
in
the
atmosphere
and
in
the
rocks
in
in
the
oceans,
especially,
and
so
you
have
this
oxygenation
event
that
occurs
and
it
kind
of
shifts.
A
You
know
the
timing
on
these
things
are
sort
of
estimated
from
a
number
of
different
sources.
So
we
have
estimates
from
molecular
clock
techniques
and
we
have
estimates
from
fossil
remains,
and
so
the
the
dates
are
not
by
any
means
like
exact.
A
Before
the
boring
billion
is
not
very
much,
what
we
have
before
four
billion
years
ago
is
a
lot
of
different.
You
know
we
have
like.
There
are
different
hypotheses
of
what
was
going
on
at
the
time.
There
could
have
been
things
like
rna
world,
which
is
where
you
have
a
bunch
of
rnas
that
are
just
freely
living
in
liquids
or
on
clay
rocks.
A
Then
you
have
this
great
oxygenation
event,
which
is
here,
and
then
you
get
up
to
this
lika
ancestor
so
lika.
This
is
actually
what
they
call
luca
down
here
four
billion
years,
and
I.
A
It
because
I
just
forgot
about
the
label,
but
that's
luca,
and
then
this
is
lika,
so
this
is
the
one.
This
is
the
least
or
the
last
eukaryotic
common
ancestor.
Then
luca
is
the
last
universal
common
ancestor.
So
you
see
that
there
is
a
universal
common
ancestor
between
bacteria
non-bacteria
and
then
there's
this
last
eukaryotic
common
ancestor
between
all
these,
what
they
call
eukaryotic
organisms,
and
so
all
of
these
organisms
have
a
number
of
different
traits
that
define
them,
and
so
from
1.5
billion
years
ago
we
have
lika.
So
we
have
this.
A
A
You
can
look
into
the
literature
on
this,
but
the
eukaryotic
cells
have
a
different
sort
of
they
have
different
properties
to
them.
So
but
we
don't
really
know
much
of
what
the
common
ancestor
looked
like,
because,
as
we
can
see
here
at
about
a
little
bit
after
the
time
that
we
had
this
last
eukaryotic
common
ancestor,
we
start
to
get
things
like
a
cell
nucleus.
C
A
Photosynthesis-
and
so
these
are
things
that
you'll
see
in
plants
and
some
you
know
some
photoplankton
and
things
like
that,
so
they're
all
you
know,
this
is
something
that
evolves
relatively
early
and,
you
know,
doesn't
exist
same
animals,
but
you
know
this.
These
are
things
that
evolve
and
be
given
the
timing
of
these
different
events.
A
It
could
very
well
be
that
this
is
around
the
same
time
as
lika,
so
the
timing
of
this
tree
might
be
off
a
little
bit
in
terms
of
what
actually
happened,
but
suffice
it
to
say
that
this
eukaryotic
common
ancestor,
you
know,
maybe
had
proto
versions
of
these
traits,
and
then
they
are
clearly
recognizable
in
the
fossil
record
around
this
time
or
in
the
you
know,
they
can
trace
back
the
origin
of
this
to
this
time.
These
these
are
sort
of
like
upper
bounds
to
the
existence
of
these
things.
A
So
we
know
that,
like,
for
example,
we
have
evidence
that
the
we
can
date
these
things
back
to
this
time,
but
they
may
always
existed
beforehand.
We
just
don't
know,
and
so
from
lika.
We
have
this
split
between
protists
and
then
we
have
this
other
split
which
leads
to
fungi
plants
and
animals.
So
we
still
have
this
very
general.
A
We
don't
really
have
a
lot
of
multicellularity
yet
down
at
lika.
We
do
get
some
multicellularity
up
further
along
here.
I
didn't
put
multicellularity
on
here,
but
what
you
do
have,
though,
is
you
have
this
quick
succession
of
events,
and
this
is
sort
of
at
the
upper
bound
of
this
born
billion
and
what
you
have
is
you
have
the
split
between
fungi
and
plants
and
animals
at
I
guess,
950
million
years
ago,
and
then
you
have
this
split
bleaching
plants
and
animals
at
850
million
years
ago.
A
A
You
know
the
order
of
these
events,
but
they
they
happened
right
around
the
same
time,
and
so
but
but
what's
interesting
is
that
you
have
this
second
oxygenation
event
which
the
first
oxygenation
event.
The
great
oxygenation
event
is,
where
you
have
this
sort
of
the
birth
or
this
sort
of
first
expression
of
oxygen
in
the
atmosphere.
A
The
second
oxygenation
event
is
more
extensive
and
the
level
of
atmospheric
oxygen
goes
up
quite
a
bit
from
its
previous
level,
and
so
this
enables
a
lot
of
innovation
in
these
eukaryotic
lineages
and
these
different
types
of
eukaryotic
life,
and
so
it's
clear
that
from
here
the
data
points
to
the
sort
of
rapid
diversification
right
around
the
second
oxygenation
event,
and
then
I
put
the
doshanto
embryo
on
here,
and
so
I
think
last
week
we
talked
about
this.
This
is
the
first
evidence
of
an
embryo
in
the
fossil
record,
so
plants
have
embryos.
A
We
haven't
talked
about
that
in
the
group,
but
plant
these
seeds
of
plants.
They
do
have
embryos
that
you
know
it's
a
bunch
of
cells.
In
an
embryo
I
had
a
professor
once
it
called
you
know
a
seed
baby
in
a
box
with
its
lunch.
It's
a
little.
A
You
know
new
places
or
to
plant
itself
in
the
earth
and
grow
and
animal
eggs
kind
of
have
the
same
function
where
they
you
know
you
have.
You
know
different
animals
that
lay
eggs,
they
can
lay
eggs
and
either
you
know,
sit
on
them
or
leave
them
and
they
hatch,
and
so
they
have
this
protective
shell
and
then
eventually
we
get
into
placental
mammals
where
they
give
live
births.
And
so
that's
later
that's
further
up
the
tree
here.
But
so
this
is
the
layout
of
the
boring.
A
A
There
is
not
a
lot
on
the
bacterial
side,
although
maybe
there's
a
lot
going
on
and
we
haven't
figured
it
out
yet,
but
I
mean
what
we're
getting
is
we're
getting
this
sort
of
you
care,
slow,
eukaryotic
and
if
you
know
maybe
some
innovations
that
are
kind
of
coming
together
and
maybe
need
to
be
enabled
by
the
section
oxygenation
event
which
then
launches
us
into
the
modern
version
of
life,
and
so
that's
that's
something
that
I'm
gonna
send
to
them
pretty
soon,
and
you
know
I
I
have
some
references
here,
so
I
didn't
just
pull
those
numbers
out
of
a
hat.
A
I
had.
I
went
through
the
literature
on
this
and
it's
not
exhaustive,
but
I
think
it
gives
it
a
broad
outline
of
it.
So
if
you're
interested
in
this,
I
can
send
you
a
link
to
this
document
and
share
it
with
you
and
I'm
thinking
of
better
ways
to
visualize
this,
but
I
don't
know
yet
might
be
better
to
yeah
you'd.
A
Access
to
it,
I'm
thinking
of
doing
like
a
circular
version
of
it,
but
I'm
still
trying
to
figure
out
how
to
space
everything
out,
but
I
think
this
is
basically
the
basic
idea
of
what
we
want
here
so
okay.
Now
I
wanted
to
get
to
the
well
one
more
thing
I
wanted
to
show
before
I
get
into
the
other
things
here
that
the
paper
is
in
that
so
there's
this
simulation
that
I
found
online.
This
is
really
nice
and
we've
talked
about
like
critical
processes
before
so.
A
This
is
it's
like
a
long
description
of
this.
It
has
like
relevance
to
viral
pandemics,
viral
epidemics,
and
so
this
is
it's
something
that
demonstrates
criticality
on
a
network,
and
so
this
is
a
long
sort
of
read,
but
it's
a
nice
interactive.
C
A
A
A
So
they
talk
about
networks
and
they
talk
about
diffusion
on
networks,
and
so
this
is
an
important
topic
for
diseases,
but
also
for
memes,
which
are
ideas,
and
you
can
study
memes
on
social
media,
the
spread
of
ideas,
a
wildfire
breaking
out
across
the
landscape,
neutrons
cascading
through
a
hunk
of
enriched
uranium.
So
there
are
a
lot
of
uses
for
a
network
diffusion
model,
and
so
he
actually
explores
this
in
an
interactive
way.
This
is
a
simple
model
of
a
network
and
in
epidemiology
which
he
focuses
on
here.
A
You
have
active
nodes
or
infected
nodes,
and
you
have
nodes
that
are
uninfected
or
not
active.
You
have
these
edges
where
by
you
can
go
from
an
infected
node
and
you
could
and
and
that
infection
can
travel
along
the
edge
of
a
certain
probability
so
that
all
these,
these
four
nodes
that
are
immediately
downstream
of
this
node
can
be
infected
with
a
certain
probability.
A
And
then,
if
this
one
says
infected,
then
it
can
spread
its
infection
to
these
four
neighbors
with
a
certain
probability,
and
so
then
it
can
spread
the
infection
through
this
network
and
it
depends
on
how
connected
each
node
is
and
the
probability
of
infection
and
that's
how
things
spread.
So
this
is
how
a
disease
spreads
through
a
population.
A
They
interact
those
people
and
they
spread
and
so
on
and
so
forth,
and
so
this
network-
you
know
it's
not
a
linear
model,
it's
it
can
happens,
maybe
in
a
linear
way,
but
it
has
non-linear
effects
and
they're
things
like
cascades
or
you
know
other
types
of
phenomena
that
are,
you
know
hard
to
predict
using
a
linear
model.
You
can
also
have
critical
phenomena.
A
So
if
you
have
this,
if
this
edge
node
say
is
infected
and
then
it
spreads
its
infection
to
a
node
that
is
highly
connected,
such
as
this
one,
it
can
infect
a
lot
of
other
nodes
very
quickly,
whereas
this
edge
node.
If
it
doesn't
infect
one
of
the
more
centrical
nodes,
then
the
infection
dies
out,
and
so
this
is,
this
is
kind
of
a
way
to
think
about
network
criticality
using
sort
of
a
an
infectious
disease
model.
A
And
so
this
is
this
is
where
they
put
the
network
on
a
grid,
and
this
is
good
for
modeling
purposes,
because
you
can
figure
out
like
the
neighbors
in
the
neighborhood.
A
And
so
this
is
one
of
the
simulations
here
he's
showing
that
an
active
node
always
transmits
its
infection
to
its
uninfected
neighbors.
In
this
way,
so
it's
sort
of
a
linear
spread
and
then
but
this
is
dull
and
then
far
more
interesting.
Things
happen
when
transmission
is
probabilistic.
So
in
this
case
it's
it's
deterministic
and
then
in
this
case
it's
going
to
be
probabilistic
here.
A
Oh
okay,
it
didn't
work
yeah,
okay,
there
we
go
see
now,
it's
it
didn't
really
transmit
much
at
all
transmission
rate
of
88
percent.
It
transmits
far
more
like
the
linear
case.
So
now
we
find
out
that
probabilities
of
infection
of
the
neighbors
actually
makes
creates
this
sort
of
asymmetric
effect
on
this
grid,
and
so
you
can
play
around
with
those
parameters
and
then
going
critical.
A
So
now
you
have
this,
you
think
about
the
longevity
of
these
things
and
transmission
rates,
and
then
this
limitless
diffusion
as
many
names
going
viral
or
going
nuclear
or
going
critical.
So
when
something
really
breaks
out
and
spreads
all
over
the
place,
it's
you
know
we
think
of
that
as
viral.
But
it's
also
going
it's
a
critical
process
and
that's
more
of
the
physics
idea
of
it.
A
You
can
have
these
things
that
are
sub-critical,
which
are
these
local
expansions,
but
a
critical
expansion
is
where
it
just
kind
of
breaks
out
and
becomes
a
global
phenomenon.
So
this
tipping
point
is
called
the
critical
threshold
and
it's
a
pretty
general
feature
of
diffusion
processes
on
regular
networks.
So
again,
this
is
an
example
of.
A
Something
where
there's
some
value
a
critical
value
here,
so
if
we
reset
it,
we
see
that
we
start
off
with
this
little
cluster
in
the
middle
and
it's
kind
of
spreading
out
and
then
at
some
point.
It
just
starts
to
spread
everywhere
and
it's
maybe
hard
to
see
where
that
critical
threshold
is,
but
you
can
see
that
it.
It
changes
its
behavior
at
some
point
and
it
becomes
a
global
phenomenon.
A
A
If
we
go
up
to
43
percent,
it's
pretty
much.
You
know
it
just
goes
global,
pretty
quickly,
81
yeah,
it's
just
linear,
just
expands
without
any
sort
of
anything
to
stop
it.
So
this
transmission
rate
then,
is
going
to
determine
how
these
things
break
up
to
being
a
critical
phenomenon
from
a
subcritical
phenomenon.
A
And
so
that's,
and
so
he
discusses
this
a
little
bit
more
in
terms
of
infection
rates
and
epidemiological
models
and
other
things,
and
then
he
talks
about
spontaneous
activation,
which
is
sort
of
like
you
know,
sort
of
a
probabilistic
thing.
But
this
is
actually
interesting
if
you're
interested
in
neurons
networks
of
neurons
and
so
the
spontaneous
activation
rate.
A
Let's
see
okay,
so
it's
not
yeah,
it
just
kind
of
appears
in
different
parts
of
the
net
in
different
parts
of
this
grid,
and
it's
you
know,
so
you
get
like
this.
These
infections
in
different
parts
of
the
grid
and
then
the
transmission
rate
will
determine
whether
these
things
take
off.
So
you
can
see
you
had
a
couple
spots
where
they
start
to
pop
up
subcritically
and
then
eventually
they
take
root,
and
then
you
get
this
expansion
to
a
global
state.
A
Okay,
so
and
then
this
I
mean
it
goes
on.
This
goes
on
for
quite
a
while,
and
so
I
didn't
want
to
spend
too
much
more
time
on
this.
But
it's
a
very,
I
think
it's
a
very
good
set
of
simulations
for
this
idea.
He
talks
about
scientific
networks
and
the
diffusion
of
knowledge,
which
is,
if
you're
interested
in,
like
even
if
you're
interested
in
the
philosophy
of
science,
history
of
science.
A
That
might
be
good
to
check
out
also
if
you're
interested
in
epidemiology
or
even
just
networks
and,
like
I
said,
if
you're
interested
in
networks
of
neurons,
it's
also
very
good
to
think
about
like
how
neurons
propagate
electrical
signals
across
their
network
anyways.
I
think
that's
a
good
pedagogical
tool
check
it
out.
I
put
the
link
in
the
chat,
and
so
the
final
thing
I
wanted
to
talk
about
today
was
this:
I
have
a
bunch
of
things
in
our
papers.
Queue.
A
First
of
all,
I'd
like
to
congratulate
my
oak
who's,
not
here
today,
but
he
just
got
accepted.
It
is
an
intern
at
run,
runway
ml,
and
so
this
is
a
company
runway
ml,
they're,
building
the
next
generation
of
creative
video
tools,
with
machine
learning
and
he'll,
be
working
on
building
the
machine,
learning
part
behind
runway
ml
and
was
thanking
people
who
are
supporting
him
on
this.
So
congratulations
mayork.
I
know
you've
been
very
good
at.
A
He
was
one
of
our
summer
code
interns
last
year
and
he
did
a
lot
of
work
with
video
in
that
project
and
he's
been
working
on
a
lot
of
gans.
So
he's
been
doing
some
things:
generative
models
of
embryos
and
other
types
of
you
know,
patterns
pattern
formation,
and
so
this
is
very
good.
I'm
glad
to
see
that
he's
finding
success.
A
So
congratulations
on
that
and
then
I
wanted
a
couple
other
things
I
wanted
to
go
through
here,
so
a
couple
papers
that
we
have
and
again,
if
you
have
to
leave
at
the
top
of
the
hour
here,
it's
fine
otherwise
we'll
go
on
for
about
15,
more
minutes
on
this.
A
So
one
of
the
papers
that
I
wanted
to
mention
here.
Okay,
we
have
actually
a
couple
things
in
the
chat
before
getting
into
this
okay.
So
I
have
a
couple:
okay,
so
for
the
periodicity
paper
there
are
papers
about
the
frequency
of
calcium
concentration.
As
an
embryo
develops,
I
was
wondering
if
calcium
periodicity
could
be
studied
alongside
cell
division
data
yeah,
probably
that
that
could
be
a
thing
that
we
could
look
at
or
mention
in
the
final
version.
A
I
think
I
don't
know
how
it
would
connect
like
directly.
A
I
know
there's
some
things
about
like
cell
cycle
and
the
timing
of
cell
cycle
in
terms
of
genetic
networks,
genetic
regulatory
networks,
but
I'm
sure
that
calcium
periodicity
probably
also
plays
a
role
in
that
as
well,
and
so
we,
I
think
we
mentioned
like
the
physics
of
cells,
so
we
we
might
bring
that
up
in
the
in
the
paper
and
then
surety
asked
is
the
transmission
rate
the
same
as
we
call
our
zero?
Yes,
yes,
that's
very
that's
very
similar
concept.
A
The
transmission
rate
is
r0
on
a
net
but
they're
just
showing
this
on
a
network
or
in
a
lattice
of
interacting
people
or
interacting
nodes
and
they're,
just
showing
this
rate
of
diffusion
in
the
in
the
population,
and
I
think
r0
is
something
specific
to
epidemiology
where
it
refers
to
the
infected
in
infectibility.
A
So
you
know
it's
like
a
combination
of
like
the
probability
of
being
infected
with
your.
Maybe
your
you
know,
assuming
a
certain
level
of
social
interaction,
so
that
that
is
very.
It
is
a
very
similar
parameter,
but
the
the
simulation
I
showed
you
largely
just
kind
of
takes
us
a
little
bit
makes
us
a
little
bit
more
abstract
and
shows
you
sort
of
the
process
behind
these.
These
things
like
how
do
things
diffuse
in
a
network?
A
You
know
it
could
be
disease,
it
could
be
information,
it
could
be
anything
you
can
think
of
electrical
signals,
whatever
the
idea,
that's
basically
a
very
similar
process
underlying
all
of
that,
so
so
the
so.
This
is
the
first
paper
I
have
for
today.
A
What
machine
learning
can
do
for
developmental
biology-
and
this
is
something
that's
very
brand
new
in
development-
and
this
is
a
paper
that,
where
they
it's
a
review
of,
basically
for
people,
non-specialists
in
machine
learning,
people
who
don't
attend
this
meeting,
you
know
developmental
biologists,
maybe
want
to
know
a
little
bit
more
about
what
machine
learning
is
all
about,
because
you
hear
about
it
and
if
you
don't
interact
with
the
literature-
and
you
don't
know
what
it's
about
so.
A
It
but
this
is
a
nice
review.
I
think
that
developmental
biology
has
grown
into
a
data
of
intensive
science
with
the
development
of
high
throughput
imaging
and
multi-omics
approaches
which
are
like
genomics,
but
they
you
know
they
have
like
phenomics
and
they
have
other
types
of
omics.
They
have
all
these
like
next
generation
sequencing
technologies,
and
so
you
can
use
all
those
you
know.
Those
are
all
part
of
developmental
biology.
Now.
A
Machine
learning
is
versatile,
set
of
techniques
that
can
help
make
sense
of
these
large
data
sets
with
minimal
human
intervention
through
tasks
such
as
image
segmentation
super
resolution
microscopy
and
cell
clustering.
In
this
spotlight,
I
introduce
these
key
concepts,
advantages
and
limitations
of
machine
learning
discuss
how
these
methods
are
being
applied
to
problems.
A
I
also
focus
on
how
machine
learning
is
improving
microscopy
in
single
cell
omics
techniques
and
data
analysis,
and
then
it
provides
an
outlook
for
the
future
of
these
fields
so
yeah
he
does
a
pretty
good.
This
is
paul
villatree
who's,
a
developmental
biologist.
I
think
he's
interested
in
computation
he's
a
friend
in
in
france.
A
I
don't
know
what
institution,
but
he
goes
over
a
lot
of
like
the
recent
developments
in
developmental
biology,
technical
developments
focusing
on
the
molecular
scale
and
cellular
scale,
and
then
he
kind
of
gets
into
the
sort
of
the
ways
in
which
machine
learning
is
applied.
So
he
talks
actually
about
some
of
the
aspects
of
machine
learning
here.
A
So
you
have,
you
know
the
things
in
developmental
biology
and
then
you
have
these
things
in
deep
learning
and
in
machine
learning
like
alphago
and
other
approaches
to
go,
which
are
games,
that
machine
learning
has
solved
or
has
done
very
well
on,
I
should
say,
and
so
then,
how
do
you
integrate
these?
A
A
You
know
for
years
using
some
computer
vision,
techniques,
various
computer
vision
techniques,
but
now
with
machine
learning.
We
can
do
this
very
quickly
on
a
lot
of
images
and
it
makes
it
a
lot
more
amenable
to
some
of
these
large
screens
that
they
do
in
developmental
biology
where
they
want
to
look
for
like
different
phenotypes
for
a
bunch
of
mutants
and
they'll.
A
It
kind
of
gives
us
his
own
take
on
like
the
hierarchies
of
theory
within
the
field
of
artificial
intelligence.
So
it's
like
there's
ai
and
then
machine
learning,
nested
inside
of
that
and
then
deep
learning
nested
inside
of
that,
so
I
mean
that's,
that's
his
approach.
Maybe
you
don't
agree
but
and
then
okay.
So
then
the
next
thing
is
image
resolution,
which
is
where
you
can
look
at
localizing
individual
molecules.
So
we
don't
haven't
talked
about
that
too
much
in
this
group.
A
But
you
know
you
can
use
different
molecular
probes,
fluorescent
probes
to
sort
of
visualize
proteins
and
gene
expression
and
other
biochemical
assay
type
things,
and
then
you
know
identify
these
using
machine
learning,
pick
them
out
of
a
background,
and
this
has
been
done
traditionally
either
through,
like
visual
inspection
or
through
some
sort
of
thresholding,
and
so
with
machine
learning.
You
can
do
this
much
much
more
effectively.
A
You
can
do
label
free
imaging,
which
is
not
labels
in
the
machine
learning
sense,
but
labels
in
like
amino
amino.
They
call
it
amino
based
staining
where
you're
staining
for
different
things
in
the
in
the
cell,
and
so
there
are
other
ways.
That's
another
way.
Another
thing
you
can
improve
upon
and
then
data
integration.
So
this
is
something
we've
been
exploring
in
this
group
where
you
can
take
different
sources
of
data
like
you
know,
different
types
of
things
like,
for
example,
you
know
we
take
like
images.
C
A
The
cell
that
are
sort
of
these
bright
field-
images
where
you
have
just
the
there
are
these
gray
images
that
we
show
with
the
cells,
and
then
we
have
these
fluorescent
images
where
they
have
a
black
background,
and
the
only
thing
you
can
see
are
these
stained
or
these
fluorescent
parts
of
the
image.
And
then
you
can
take
those
two
sets
of
images
and
merge
them
so
that
you
have
information.
You
can
co-register,
essentially
the
cell,
the
geometry
of
the
cell,
with
some
marker
that
you're
looking
for
and
put
it
into
a
spatial
context.
A
A
We
can
use
this
approach.
You
know,
then
we
can
use
machine
learning
on
single
cellomics.
So
this
is
where
we're
looking
at
rna
sequencing
in
single
cells,
3d
confirmation
of
dna
of
quantity,
of
transcripts,
which
are
like
a
transcriptional
output
of
of
different
for
different
genes.
A
You
can
use
you
can
use
machine
learning
on
these
data
because
these
methods
generate
huge
data
sets,
and
so
you
need
a
way
to
like
get
through
this
data
quickly.
You
know,
and
and
and
and
machine
learning
lends
itself
to
that
now.
We've
talked
about
things
like
t-sne
and
new
map,
and
he
mentions
them
here
under
clustering.
So
these
are
methods
that-
and
I
know,
if
you
many
some
of
you,
didn't
see
this
talk.
I
gave
it
a
last
december
on
t-sne
and
umap.
A
So
that's
something
that
if
you
want
to
look
at
that
presentation,
those
are
different
methods
for
sort
of
clustering
data
a
lot
of
times,
if
they're
doing,
building,
building
up
a
transcriptional
profile
of
a
single
cell,
for
example,
they'll
cluster,
the
transcripts
in
different
ways,
and
you
can
visualize
them
and
that
machine
learning
helps
in
that
process.
A
Then
he
gets
into
deep
learning
and
the
inference
of
spatial
and
temporal
relationships
which
we
talked
about
and
then
he
talks
about
growing,
an
interdisciplinary
community.
So
this
is
something
we
might
revisit
because
they
talk
about
how
to
foster
this
interaction
between,
like
computational
methods,
machine
learning
and
the
development
of
biology
itself.
And
it's
not
easy
because
you
know
our
group
is
relatively
unique
in
in
the
world
of
this.
A
It's
a
very
small
group
of
people
who
are
sort
of
doing
these
things
in
parallel,
and
so,
but
we
might
try
to
think
more
in
terms
of
community
as
well
and
so
yeah,
and
then
he
talks
about
storing
data
computational
resources.
A
And
if
you
want
to
read
it
more
about
it,
you
can
I'll
give
you
the
link
to
the
drive,
put
it
in
the
chat,
so
ask
for
permissions
for
that.
If
you
need
it,
and
so
that's
that's
one
paper.
Another
paper
that
I
ran
across
this
week
was
this:
now
we're
shifting
back
to
developmental
biology
and
actually
embryology,
and
this
is
a
historical
profile.
A
C
A
He
remembered
at
best
as
a
shadowy
figure
among
those
who
gradually
built
up
our
current
view
of
evolution
and
the
role
of
genetics.
This
view
derives
from
the
modern
synthesis
recognition
that
emerged
in
the
30s
1930s,
that
genetics
can
adequately
explain:
darwinian
evolution
and
speciation
through
natural
selection.
A
So
before,
like
the
1930s,
they
didn't
really
have
a
good
genetic
basis
for
natural
selection
and
speciation,
but
in
the
modern
synthesis
which
was
kind
of
formed
in
the
30s
people
sort
of
had
they
were
starting
to
build
up
a
foundation
a
genetic
foundation.
A
For
this
I
argue
that
de
beers
theories
of
embryology
had
a
crucial
role
in
the
modern
synthesis
and
that
his
work
indirectly
continues
to
influence
how
we
think
about
the
genome,
evolution
and
developmental
biology,
and
so
so
he
was
born
in
the
uk
and
he
was
educated
in
paris
and,
let's
see
he,
his
knowledge
was
in
seek
encyclopedic.
A
He
was
a
gifted
linguist
he's,
especially
interested
in
darwin.
Let
me
get
to
the
relevant
stuff
here.
So.
C
A
A
So
if
you
look
between
lizards,
tortoises,
pigs
and
humans,
you
see
that
in
there's
a
certain
stage
of
you
know
embryogenesis,
where
these
all
look
very
similar
right,
despite
how
different
they
are
in
their
sort
of
final
form
or
their
adult
form.
They
look
very
similar
here
and
then
eventually
they
start
to
differentiate
as
as
embryogenesis
proceeds,
and
so
when
you
don't
have
a
genetic
basis
for
that,
you
think
well,
there's
there
must
be.
You
know
they
must
all
be
basically
from
the
same
sort
of
origin,
and
then
they
have
this
developmental
divergence
here.
A
Like
from
the
egg
and
that
they
get
expressed
sequentially
and
that,
even
though
they
look
the
same
here,
they
have
different
genomes
and
they're,
not
necessarily
from
the
same
origin,
so
hakel
sort
of
had
this
idea
in
the
19th
century
before
we
had
genes
or
this
concept
of
genes
and
gene
expression
and
all
this,
and
so
this
is
the
position
that
was
widely
held
actually
this
time,
and
so
the
concept
of
recapitulation
was
something
that
they
used
to
think.
This
was
what
this
this
viewpoint
of
the
embryo
was.
A
Is
that
development
recapitulates
a
lot
of
this
sort
of
so
the
recapitulation
is
like
where
they
all
basically
have
the
same
shape,
and
then
they
diverge
out.
But
of
course
you
know
they
didn't
understand
the
mechanism
underlying
it,
so
they
thought
of
it
as
sort
of
like
a
you
know,
a
natural
stage
in
every
organism,
and
it
wasn't,
there
was
no.
There
was
no
underlying
theory
of
of
what
was
going
on,
so
they
were.
A
They
were
misguided
in
that
view,
so
basically,
this
person
helped
to
correct
this
idea
and
ground
it
more
in
genetics.
So
now
we
understand
why
you
know
you
know
you,
you
start
off
with
an
egg.
You
start
off
with
these
very
similar
forms
and
you
end
up
with
likely
divergent
forms
and
that
it
isn't
just
because
they're
all
kind
of
from
the
same
origin.
We
know
that,
like
there
are
a
lot
of
things
going
on
underneath
the
hood
here,
so
so,
dehazel
embryogenesis
embryogenesis
had
a
solely
phyllogenetic
meaning.
A
This
notion
was
captured
in
the
catchphrase
ontogeny
recapitulates
phylogeny,
the
scheme
he
proposed
was
lamarckian,
so
lamarck
was
someone
who
believed
that
behavior
influenced
evolution,
and
that
was
incorrect
in
terms
of
the
darwinian
framework
and
so
in
the
marquee
in
evolution.
Evolutionary
novelties
can
only
be
acquired
at
the
adult
end
of
development,
because
earlier
stages
of
embryogenesis
were
thought
of
to
be
fixed,
the
fixed
adult
forms
of
ancestral
organisms,
and
so
must
be
invariant.
A
So
there's
a
lot
of
you
know
this.
This
article
really
goes
in,
I
think,
into
the
historical
aspects
of
like
how
people
used
to
think
about
evolution
and
development,
and
now
that
really
changed
over
time
over
about
200
years
and
there
were
a
lot
of
ideas
in
early
embryogenesis
that
were
you
know,
and
this
is
true
of
any
science,
where
you
have
early
ideas
that
are
incorrect,
not
because
people
were
stupid,
then,
but
because
they
didn't
have
like
all
the
information
that
they
would
need
to
understand
the
system.
So
you
know.
A
About
genomics,
if
you
don't
know
anything
about,
like
you
know
some
of
the
underlying
biochemistry
of
organisms
then
or
even
heredity,
then
if
you
don't
have
those
theories
and
ideas
in
place,
then
it's
really
hard
to
make
the
correct
interpretation.
You
make
interpretations
that
seem
correct,
and
this
may
hold
true
even
for
what
we
know
today.
We
may
not
really
have
a
full
appreciation
of
what's
going
on
in
development,
even
given
our
knowledge
of
the
genome.
So
these
are
some
nice
figures.
A
If
you
want
to
look
this
paper
over,
there
are
a
lot
of
nice
figures
that
show
this
these
sort
of
processes.
So
there's
the
pre-formation
hypothesis
there's
the
one-step
transformation
hypothesis,
there's
genetic
pre-formation
and
there's
epigenesis,
which
is
currently
the
orthodoxy
and
and
makes
sense
given
what
we
know
today.
So
epigenetic
epigenesis
is
where
you
have
you
start
off
with
this
bag
of
genes
and
a
single
cell,
and
then
you
get
differentiation
of
genes
or
of
cells
and
you
get
changes
in
shape,
and
then
you
end
up
with
this
frog
before
that.
A
So
now
he
gives
an
example
of
how
different
thinkers
sort
of
thought
about
development,
so
von
bere,
who
preceded
hakel,
thought
of
development
in
a
certain
way,
and
he
had
certain
assumptions
and
then
hakel
came
along
and
changed
that
a
bit,
but
it
was
still
incorrect
at
least
the
way
we
understand
it
today
and
then
now
we
have
this
revised
version,
which
is
where
we're
thinking
about
this
explicitly
in
a
phylogenetic
context.
Although
hecal
had
a
phylogenetic.
A
Approach
as
well,
it
was
based
on
these
changes
happening
as
you
go
along
the
span
of
development,
so
that
every
embryo
was
basically
from
the
same
sort
of
source
material
and
then
got
shaped
in
development,
whereas
nowadays
we
know
that
there's
this
evolution
of
development,
where
development
evolution
constrains
development
and
then
there
changes
in
development
but
they're
not
due
to
like
this
sort
of
preformed
view
of
the
world's
or
preformed
view
of
of
organisms.
So
it's
it's.
You
know
it's
it's
there's!
A
Some
of
these
differences
are
very
subtle,
but
they're
important
to
recognize,
because
you
know
when
you
put
together
a
theoretical
argument.
Sometimes
it's
hard
to
you
know,
make
these
distinctions
and
understand
the
history
of
all
this.
So
I
would
definitely
take
a
look
at
this
paper
if
you're
interested
in
the
sort
of
the
history
of
biology,
but
also
the
history
of
how
theories
are
formed
and
and
other
things,
and
so
this
also
brings
us
to
this
era
of
genomics
in
that
in
our
common,
our
current
era.
A
A
Homology
and
unsolved
problems
de
beers
summed
up
these
issues
in
a
mastery
fashion.
Here
the
genes
are
no
more
reliable
than
adults
or
embryological
characters
as
measures
of
homology,
which
is
like
the
similarity
between
different
species
and
what
applies
to
embryos
and
adults
must
apply
to
the
genes
that
determine
them
with
the
admin
and
molecular
genetics.
A
So
I
think
that's
good.
Let
me
look
in
the
chat
again,
so
citizens
have
mechanically
similar
processes
vertebrates.
Okay,
so
I
mean
that's
yeah,
so
that's
yeah!
So
in
there
you
know
we,
if
you
want
to
if
people
want
to
follow
up
on
some
of
these
papers,
that
I've
shown
it's
great
or
some
of
the
issues
that
we
talked
about
today.
A
That's
great
so
yeah,
that's,
I
think
that's
all
I
had
for
today.
Thank
you
for
meeting
next
week.
Maybe
we'll
talk
more
about
some
of
these
other
submissions.
I
actually
didn't
even
talk
about
the
flash
talk
for
this
osf
conference.
I
didn't
get
into
those
slides.
I
don't
really
have
them
organized
yet,
but
what
we
need
to
do
for
this
is,
I
think,
create
a
five
to
ten
minute
talk
on
divo,
learn
and
even
learn.
A
You
know
the
platform
and
how
it
might
be
used
in
education
and-
and
you
know
getting
our
ideas
across
in
the
group,
so
I
think
that's
something
that
we'll
do
next
week,
I'm
going
to
work
on
it.
Some
the
presentation
for
that
is
on
february,
7th
and
I'll-
send
some
more
information
about
that
event,
closer
to
the
date
they're
going
to
have
a
lot
of
stuff
on
online
education.
A
So
if
you're
interested
in
that,
you
might
want
to
attend
the
unconference
it's
it's
on.
I
think
it's
on
like
zoom
or
something
like
that
and
I'll
I'll
get
the
slides
in
better
shape
for
next
week
and
we'll
go
through
them.
C
A
Followed
up
on
the
vertebrates
developer
on
a
beating
heart,
so
this
is
this
had
to
do
with
the
paper
that
we
talked
about
yeah
and
so
susan.
If
you
have
any
papers
that
you
know
of
that,
might
be
relevant
to
following
up
on
this
paper
or
some
of
the
other
stuff,
why
don't
you
send
them
along
in
an
email
and
I'll
we'll
do
them
next
week?