►
From YouTube: DevoWorm (2020, Meeting 21): GSoC updates, papers on beak/wing shape, "Periodicity in the Embryo"
Description
Attendees: Bradly Alicea, Mayukh Deb, Jesse Parent, Ujjwal Singh, Richard Gordon, Abhishek Bvs, and Vinay Varma.
A
B
B
C
A
All
right
welcome
to
the
meeting
so
today
we're
gonna
have
updates
on
g-shock
and
we'll
have
maybe
some
papers,
and
maybe
we'll
have
a
presentation
after
that,
so
but
always
start
with
once
the
starters
will
are
my
yoke
under
weekly
progress.
Oh.
E
E
B
A
A
A
A
B
B
A
E
A
E
F
A
A
A
So
we
have
the
diva
worm
open
worm
to
curriculum,
which
has
a
number
of
lessons
in
it.
Then
we
have
the
evil
worm
ml,
which
is
the
series
on
ml
and
so
I
think
those
two
would
be,
and
those
are
both
on
the
github
in
the
github
repo
under
those
names
respectively.
So
you
know
maybe
just
link
to
those
things
and
maybe
give
a
little
description
about
what
they
are.
We
can
you
know
we
can
go
over
the
copy
on
that
when
that
comes
up
just
you
know,
when
you
bring
my
attention
to
it
over.
E
E
F
A
A
A
A
F
A
In
ji-suk
2020
we're
in
the
project
board.
Basically
you
know
if
you
want
to
share
some
or
share
some
feedback.
You
maybe
want
to
become
a
collaborator
on
this
board,
so
I
can
invite
you,
and
if
you
are,
then
you
can
put
in
a
comment
here.
This
is
a
design
evaluation
for
as
well
as
layout.
This
is
a
generic
comment
and
it
has
this
label
open,
Depot
sell.
A
So
if
you
look
and
github,
the
labels
are
here
we're
putting
this
label
which
I
made
on
those
issues,
and
then
we
can
go
through
them
systematically
in
future
meetings.
So
if
you
have
an
idea,
I
mean
you
can
send
me.
The
idea
for
like
an
update
and
I
can
I
can
put
it
up
as
an
issue
or
you
can
do
it
and
then
we'll
have
them
all
under
that
tab,
so
you
all
can
actually
look
at
for
those
you
can
search
for
those
labels
and
see
what
the
feedback
looks
like
in
terms
of
feedback.
A
I
mean
the
only
thing
I'm
still
thinking
about
is
like
the
visual
elements
issue,
which
is
like
where
we
have
the
all
those
flash
pages.
In
the
background
and
they're
like
different
colors
and
I,
don't
know
if
that's
really
distracting
or
not
to
people
or
if
it's
a
little
incoherent
visually
I
mean
I.
Think
maybe
a
solution
is
to
make
it
like
a
uniform
sort
of
mask
of
a
certain
color.
So
like
a
very
transparent
light,
green
or
something
might
be.
E
A
E
D
A
I,
just
like
kind
of
like
it's
part
of
the
template
of
the
the
site
week,
it's
part
of
the
color
scheme,
but
it
isn't
like
this:
it
doesn't,
it
isn't
jarring.
You
know,
I
mean
otherwise
I
think
it's
it's
good
to
have
like
different
backgrounds,
coming
up
with
different
topics
and
I
like
the
way
it's
laid
out
with
our
next
yeah
yeah
and
then
we'll
talk
about
the
copy
like
you
know,
for
like
the
Academy,
and
then
we
can
talk
about
how
to
set
up
the
data
like
I.
Think
for
the
data.
A
B
B
B
B
B
B
A
A
B
A
B
B
A
B
A
Yeah
I
was
trying
to
think
of
other
ways.
You
could
do
this
to
improve
the
boost.
The
performance
I
think
those
are
the
two
best
ways.
I
mean
there
might
be
well
with
ResNet,
though
you
don't
have.
You
can't
necessarily
use
adversarial
data,
which
I
mean
might
be
a
strategy,
but
I
don't
know
if
that
might
be
overkill
for
want
to
do
here,
because
it's
you
know,
what's
the
adversary.
A
B
B
B
B
B
A
A
F
F
F
E
E
E
E
B
F
A
Let's
see
we
have
yes,
he
had
a
comment.
I
was
going
to
look
at
some
deform
Academy
actually,
but
I'll
have
more
to
say
on
slack
limited
discussion,
ability
right
now
so
yeah,
please
look
over
the
stuff,
like
the
deform,
Academy
I
think
is
just
the
diva
warm
ml
and
the
evil
arm
open
worm,
curriculum
materials.
So
please
do
and
Vinay
says:
hey
goes
well.
Can
you
please
ping
the
website
link
of
its
hosted
somewhere,
so
I
think
right
now,
which
wall
has
it
like
in
a
private?
A
Jesse
says:
is
there
anything
in
the
Academy
specifically
about
machine
learning,
evil
worm
ml
I
mean
we
just
have
like
the
ml
like
the
ml
course
materials?
I,
don't
know.
If
there's
anything
more
specific
than
that
I
mean
we
can
create
materials,
but
yeah
I
mean
it's.
It's
a
little
bit
like
right
now,
yeah,
it's
a
little
bit
uncoordinated
yet
as
to
how
that's
gonna
fit
together.
A
A
You
think
especially
well
I
mean
my
hope
has
like
looks
like
he's.
He's
also
done
some
work
on
the
movement
aspect.
I,
don't
know
if
he
showed
that
part
of
it
in
the
meeting.
His
notebook
on
movement
cell
movement
I
think
that
wasn't
on
a
Friday
meeting,
one
of
the
Friday
meetings,
yeah
all
right.
So
in
coming
weeks,
where
we
revisit
that
we'll
revisit
the
website,
of
course.
So
next
thing
I
want
to
talk
about,
is
I,
guess
I'll
go
to
the
papers
for
this
week,
see
if
I
can
share
my
screen.
A
A
A
A
And
then
you
could
have
like
these
worked
coordinate
systems
that
would
describe
the
difference
between
the
morphology
of
the
different
species
and
so
I
bring
that
up,
because
in
this
paper
they
address
the
issue
of
beak
shape
diversity.
Using
this
comparative
morphometrics
analysis,
which
is
where
they
look
at
the
morphology,
they
measure
it
with
some
metrics
and
then
they
analyze
the
differences.
A
They
analyzed
beak
shapes
in
a
diverse
group
of
songbirds.
Here
we
show
that
the
dynamics
of
proliferative
the
deliberative
Grosjean,
which
is
a
part
of
the
beak
in
development
that
you
know,
controls
how
the
cells
you
know
sort
of
form.
The
beak
must
follow
restrictive
rules
to
explain
the
observed
variation
with
big
diversity,
constrain
to
a
three
parameter,
family
of
shapes
per
parameterised
by
length,
depth
and
degree
of
shear.
So
they're
talking
about
the
length
of
the
beak,
the
depth
of
the
beak
and
then
the
degree
of
shear,
which
is
a
mechanical
property.
A
We
experimentally
verify
these
predictions
by
analyzing
cell
proliferation
and
the
development
of
genetic
beat
speaks
of
a
zebra
finch
comment
here:
okay,
Jesse
I
mentioned
the
Thompson
forms
book
to
Anson
re
some
of
his
development
of
his
ideas
and
mentioned
he
come
here,
but
he
wasn't.
That's
okay
can
check
us
out
on
YouTube
or
we
can
talk
about
it
in
our
meeting.
We
are
talking
about
that
with
Anselm's.
In
my
other
group,
I
was
different,
different
thing,
okay,
so
yeah.
A
So
our
findings
indicate
the
beak
shape
variability
in
many
songbirds,
strongly
constrained
by
shared
properties
of
the
developmental
program
controlling
the
growth.
So
let's
go
down
to
the
figures,
because
those
are
often
more
enlightening
than
just
reading
text
so
well,
first
of
all,
they
they
have
their.
They
talk
about
their
they
introduce
this,
and
then
they
talk
about
the
results.
A
So
one
of
the
things
they
do
is
they
quantify
this
beak
shape
diversity,
and
one
of
the
things
that
are
doing,
of
course,
is
they're
using
a
mathematical
model
to
represent
the
beat
shape,
and
so,
but
what's
important
here
is
it
just
the
character?
Is
the
beat
shape
it's
the
character
as
a
difference
in
B,
cubes
and
so
they've
been
able
to
figure
out
the
most
important
weeks,
eight
parameters,
depth
length
and
curvature,
and
they
can
study
this
from
two-dimensional
profile
images.
A
A
These
transformations,
where
they
have
si
they're,
basically
doing
a
bunch
of
mathematical
transformations
to
find
the
difference,
they're,
finding
a
derivative
profile
and
then
they're
finding
they're
actually
using
like
I,
guess
a
more
flow
space
approach
to
understand
them.
Sort
of
how
closely
the
two
shapes
are
related.
So.
A
They're
mapping
these
shapes
and
the
parameters
to
a
space,
that's
sort
of
like
you
know,
sort
of
like
gradient
descent,
but
it's
you
know
it's
looking
for
how
closely
they're
related,
so
you
don't
want
to
just
compare
them
sort
of
discreetly.
You
want
to
put
it
into
US,
coordinate
space.
That
says
you
know.
All
these
parameters
are
related
is
like
maybe
a
three
dimensional
space.
Now,
if
we
have
this
as
a
three
dimensional
space,
where
do
our
beaks
follow
up?
They
fall
out
in
different
points
that
we
can
then
find
the
different.
A
You
know
that's
a
space,
that's
differentiable,
so
you
have
the
three-dimensional
space.
You
know,
one
dimension
might
be
beat
shape
and
you'll
have
like
a
range
of
values,
and
then
you
know,
beats
will
be
distributed
along
that
set
of
that
that
distribution
in
the
beat
shapes
will
be
within
that
distribution
and
there'll
be
two
other
dimensions,
and
then
you
can
just
simply
make
calculations
between
the
TV
chips
and
it
doesn't
necessarily
matter
if
they're,
how
closely
they're
related
in
evolution.
It's
just
that
those
relationships
are
purely
quantitative
based
on
the
data.
So
this
says
nothing.
A
This
says
a
little
bit
about
evolution,
but
it
isn't
like
it
doesn't
a
surrogate
for
evolution.
In
other
words,
if
there's
a
change
in
beat
shape
that
doesn't
say
that,
like
evolution,
you
know
it
doesn't
give
you
an
evolutionary
distance.
Necessarily,
it
just
tells
you
that
there
was
so
more
for
generous
differential
morphogenesis
in
these
two
beats
shapes
that
could
be
due
to
developmental
pressures
that
could
be
due
to
in
evolution,
but
we
don't
really
know
so.
A
A
You
know
basically
forms
of
triangle
and
they're
characterizing,
these
different
clades,
so
these
are
groups
of
species
that
are
evolutionarily
related
that
they
can
take,
and
they
can
say
this
is
the
beak
shape
for
this
group
of
species,
so
the
phylogenetic
and
morphometric
structure
of
each
shapes-
and
so
let's
see
this
figure
is
actually
and
then
they
they
basically
talk
about
the
different
parameters,
see
the
hierarchical
collapse
of
beak
shapes
unto
each
other
under
scaling
and
shear.
So
they
have.
They
do
these
pairwise
comparisons
and
beak
shapes
across
phylogeny.
A
So
all
the
beak
shapes
are
compared
pairwise
and
then
they
say
heatmap
of
e
sub
s
resulting
from
pairwise
comparisons
of
all
beat
shapes
in
the
phylogeny
scaling.
One
beat
shape
onto
another
crosses
indicate
pairs
of
species
of
beaks.
Do
not
collapse,
buying
scaling
transformations
as
there
is
no
minimum
e
sub
s
as
a
function
of
the
scaling
factors,
so
I
think
they're
looking
at
the
ones.
It's
like
the
big
shapes
that
scale
together.
A
A
They
might
be
different
shapes,
but
they
grow
in
the
same
way.
And
then
you
have
that
as
a
group
of
beaks
and
then
you
look
at
share
plus
scaling,
so
you
have
like
this
growth
scaling
and
then
you
have
this
shear
scaling
and
then
you
evaluate
them
together,
and
so
what
they
say
here
is
on
small
phylogenetic
scales,
big
shapes
collapse
under
scaling
alone.
For
example,
the
two
gs
bisa
species
on
the
top
left,
creating
groups
of
similar
beat
shapes
represented
by
colors.
A
These
group
shapes
in
turn
collapse
onto
each
other
under
shear
in
their
length
direction.
So,
specifically,
all
group
shapes
collapse
under
the
shape
of
the
blue
colored
group,
so
they
all
basically
have
some
commonality.
Under
this
blue
group,
the
blue
colored
group
can
be
approximated
to
an
extremely
high
precision
as
a
section
of
a
parabola,
as
shown
on
the
right.
That's
what
they
show
this
parabola
here.
A
The
combination
of
this
hierarchical
collapse
under
scaling
and
shear
under
the
blue,
colored
group
and
the
collapse
of
the
blue
coloured
group
under
parabola
leads
the
conclusion
that
all
beat
shapes
are
considered
as
conic
sections,
so
that
they've
done
is
they've
taken
all
the
beat
shapes
and
they're
explaining
in
a
different
number
of
different
ways.
They're
saying
that
there's
the
scaling
commonality
amongst
certain
beaks
there's
a
share
commonality,
in
other
words
they
have
common.
A
They
they
have
related
attributes
in
terms
of
scaling
in
terms
of
shear,
and
then
this
parabola,
which
explains
the
rest
of
the
variation
and
so
this
base.
Those
three
factors
will
explain:
explain
why
there
are
so
many
different
types
of
beat
shapes.
They
all
come
from
a
developmental
precursor,
there's
a
lot
of
variation,
and
these
are
three
reasons
why
you
have.
A
So
this
is
where
basically,
a
lot
of
the
growth
is
happening,
so
you
can
see
an
e5.
The
growth
zone
is
rather
large,
east,
seven,
it's
a
little
smaller
and
then
a
nine.
It's
very
small,
and
this
is
where
a
lot
of
the
cells
proliferate
and
this
growth
zone,
so
the
cells
are
just
growing
everywhere.
They're
growing
out
of
these
disrupts
on
and
then
they're
migrating
out
to
the
edges
and
the
shape
is
being
influenced
in
that
way,
so
the
shape
is
modulated
throughout
several
developmental
stages.
A
In
terms
of
how
its
you
know,
extruding
from
that
little
bump
here-
and
so
you
know,
this
means
that
you
know
given
different
species
given
different
variants
of
genes
given
different
chemical
factors,
this
can
be
controlled
in
different
species.
So,
let's
that's.
That's
probably
enough
for
that
paper.
A
A
And
there's
a
lot
of
methods
in
it
where
we
know
what
they're
did?
Will
the
big
shape
analysis
is
here
and
then
yeah
so
I'm
not
really
sure
how
to
answer
that
question,
but
I
am
going
to
share
the
folder
I
can
share
the
folder
here
in
the
chat
and
then
we
can
people
want
to
look
up
the
papers
in
more
detail.
A
There's
another
paper
that
I
have
here:
Evo
4
becomes
2,
evo-devo
4
becomes
2,
so
this
one
is
actually.
This
is
a
dispatch
from
Current
Biology,
and
this
is
actually
quite
a
different
area.
This
is
so
the
abstract
is
wings
and
halters
or
homologous
flight
appendages.
All
tiers
are
in
insects,
flight
appendages,
as
you
shape,
differences
are
controlled
by
the
EBX
transcription
factor.
Recent
research
shows
ie
BX
regulates
apical
and
basal
extracellular,
matrix
Proteus
and
their
inhibitors
to
achieve
this
morphological
divergence.
So
that's
a
lot
of
jargon,
I
guess.
A
Ruby's
research
analyze
tell
the
UV
XR
ultrabithorax
ox
protein
controls,
patterning
cell
proliferation
and
cell
differentiation,
and
to
regulate
the
differential
development
of
wing
versus
halt
here.
A
recent
paper
and
development
now
completes
the
study
by
analyzing
how
ubx
modulation
of
the
apical
and
basal
extracellular
matrix
affects
their
morphogenesis
during
metamorphosis,
and
so
this
is.
This
is
one
of
these
papers
where
they
feature
a
couple
of
papers
that
are,
you
know,
violated
in
that
issue.
So
this
is
an
example
here.
This
is
a
graphical
summary
of
the
paper.
A
A
A
A
Let's
see
so
we're
going
to
do
this
paper
a
long
time
ago
on
periodicity
in
the
embryo,
and
so
is
on
this
idea
of
differentiation
waves,
which
are
waves
in
and
I,
know.
Dick
is
involved
in
this,
at
least
at
some
level.
I
wasn't
sure,
but
I
got
the
invitation
of
my
email.
It's
about
the
idea,
there's
waves
and
development
that
contribute
to
differentiation,
and
so
it's
you
know
it's
they
just
send
out.
These
calls
for
papers
and
people
have
to
contribute
papers.
A
A
It's
called
periodicity
in
the
embryo,
and
so
this
was
about
sort
of
looking
at
the
emergence
of
order
and
space
diffusion
in
the
order
of
time.
So
the
idea
is
that
there's
a
tempo
and
mode
of
embryonic
development
and
that
you
know
cell
division
and
cell
differentiation
have
these
different
timescales,
so
I
have
a
bunch
of
data
from
different
model
systems
here
that
collected
for
this
created
graphs.
A
So
in
this
case
this
is
cell
per
zebrafish
embryo
cells
during
embryogenesis-
and
you
can
see
there's
this
there,
these
bursts
of
cell
birth
and
then
eventually,
this
number
of
the
distribution
of
cell
births
becomes
more
diffuse
over
time.
But
you
know
you
can
get
a
sense
of
maybe
what's
going
on
underneath
the
hood
by
looking
at
these
patterns.
This
is
from
cell
tracking
data
for
zebrafish,
there's
also
corresponding
data
from
C
elegans,
where
you
have
cell
burbs
and
development,
and
so
it's
a
little
harder
to
see.
A
There
are
these
bursts
of
activity
as
well,
and
so,
if
we
go
to
the
abstract,
we
talked
about
looking
at
sort
of
these
patterns
and
time
and
perhaps
in
space
using
two
model
organisms,
and
so
we
had
the
data
from
the
zebrafish
and
from
C
elegans
and
are
both
based
on
cell
tracking
data
from
development.
Where
you
can,
you
know,
track
the
birth
of
birth
times
of
cells
and
figure
out,
like
these
patterns
of
cell
division
and
differentiation,
and
so
based
on
these
findings.
A
We
predict
that
periodic
changes
in
the
frequency
of
new
cells
over
developmental
time
represents
cell
proliferation,
the
functional
distinction,
a
periodic
changes
in
sub
birth
frequency,
represent
a
diversity
of
specialized
sub
lineages
and
then
determine
whether
a
periodicity
represents
meaningful
information
or
noise.
We
can
use
infer
the
information
isometry
technique,
so
that's
this
graph
here,
so
this
would
be
an
application
of
this
method.
This
is
something
that
we've
worked
on
in
the
group
in
years
past,
where
we
create
these,
generate
these
plots
of
developmental
cells
from
trees
and
look
at
their
patterns
and
I.
A
Don't
know
if
that's
really
suitable
for
this
special
issue,
but
we
can
talk
about
it
in
offline,
but
I
just
wanted
to
bring
people's
attention
to
it.
Something
and
in
thinking
about
in
terms
of
a
different
you
know
so
I
knew
like
we
need
a
new
project,
but
I
mean
you
know
some
some
goal
to
shoot
for
here.
So
we
have
this.
So
that's
that's.
If
you're
interested
in
that,
let
me
know
I
think
dick
left.
The
meeting
so
I
think
we're
pretty
close
to
the
end
of
the
meeting,
so
they
stopped
presenting.
A
We
will
be
meeting
on
Friday
if
you
wanna,
if
as
well,
one
might
want
to
drop
by
and
give
a
little
update
and
their
projects
I
might
be
able
to
see
them,
but
we
can
stay
in
touch
on
slack
and
that
should
be
good
and
then
we'll
meet
the
following
week
and
then,
if
anyone's
interested
in
presenting
on
anything,
let
me
know
we
have
a
lot
of
spots.
Open,
I
know
we
give
like
the
presentations
on
the
first
30
minutes,
but
or
we
give
an
update
in
the
first
30
minutes,
but
there's
always
space
afterwards.