►
From YouTube: CI WG Demo, June Presentation
Description
WG: DSCI
Date: 06/05/20
Presenter 1: Peter Rose & Ilya Zaslavsky
Institution: San Diego Supercomputer Center
Title: COVID-19-Net – using knowledge graphs to make sense of heterogeneous COVID datasets
West Big Data Hub
Presenter 2: Florence Hudson
Institution: Columbia University
Title: COVID Information Commons
Northeast Big Data Hub
Presenter 3: Christine Kirkpatrick
Institution: San Diego Supercomputer Center
Title: Virus Outbreak Data Network, with a spotlight on VODAN Africa
West Big Data Hub
A
I'm
going
to
try
just
because
I
know
we
have.
You
know
three
lightning
talks
of
five
minutes
each
and
that
never
I'm
not
blaming
anybody,
but
it's
really
difficult
to
give
a
five-minute,
lightning
talk
and
and
keep
everybody
on
time.
So
I'm
hoping
we
can
sort
of
get
through
the
beginning
parts
here.
I
think
I
recognize
most
everybody
here
just
as
an
introduction
in
case
I'm
now
gaffney,
I'm
director
of
data
computing
attack,
and
I
co-chair
this
with
many
of
the
other
folks
here
in
the
room.
A
Every
month
we
do
a
nice
set
of
presentations
around
data,
but
then
also
talk
about
what's
going
on
in
the
hubs,
and
so
I
don't
think
we
need
to
do
around
the
room,
but
we
can
go
ahead
and
get
started
on.
If
there's
what
news
there
is
coming
from
the
hubs,
how
are
things
going?
How's
how's,
how's
tricks,
so
christine
you
want
to
start
with
the
last
real
quick.
B
Sure,
well,
I
can
start
with
what
happened
most
recently.
So
yes-
and
actually
this
is
perfect-
we're
on
the
call-
because
I
it'll
save
me
an
email
so
yesterday
and
today,
actually
at
the
same
time
as
this
call
sarah
stone
is
covering.
Today,
we
in
the
west
participated
in
the
hsi
stem
hub
virtual
residency
on
grantsmanship
hsi
stem.
You
know
what
stem
stands
for.
Hsi
stands
for
hispanic
serving
institutions,
they're
an
nsf
funded
initiative
similar
to
the
hubs,
except
they
don't
have
they're
not
divided
by
census
region.
B
It's
about
what
we
would
give
to
one
hub
covering
the
whole
us,
and
there
are
a
lot
of
hsis,
oh
yeah,
and
so
we
have
partnered
with
the
hsi
stem
hub.
This
is
we
followed
up
from
an
event
we
did
in
september,
which
was
two
days
on:
how
do
you
teach
data
science
and
and
doing
things
with
data
to
hsis
a
lot
of
community
college
instructors
and
then
teaching
college
instructors
as
well,
and
it
was
so
rewarding?
We
said,
let's
keep
working
together,
so
long
story
long.
B
We
had
our
virtual
workshop
yesterday
and
today
and
most
of
the
participants
this
time
around
are
in
the
other
hubs,
the
south
and
the
northeast,
and
so
I
have
florence
one
person
who
I
wanted
to
put
together
an
email
with
you
and
katie
just
to
be
in
touch.
She
wants
to
partner
on
some
outreach
stuff
for
an
upcoming
proposal
and
then
for
the
south.
I
have
two
people
who
wanted
an
and
so
I'll
be
doing
I'll,
be
doing
that
as
well.
B
I
hope
that's
okay
and
I
couldn't
remember,
didn't
we
say:
puerto
rico
belongs
to
everybody,
it's
it's
not
just
that
the
south
is
closest,
or
I
I'm
pretty
sure
at
the
beginning.
Puerto
rico
could
work
with
anyone,
but
I'm
still
going
to
introduce
the
the
puerto
rico
crowd
to
the
south
hub.
Thank
you.
C
A
And
what
may
be
an
interesting
future,
I
don't
know,
do
lightning
talk
or
whether
we
could
do
a
little.
You
know
group
workshop
or
something
in
this,
but
you
know
talking
about
how
the
the
hubs
can
help
support
these
these
folks
getting
crossover
between
them,
because
you
know,
there's
there's
a
lot
of
that.
That
is
sort
of
needed
and
I'd
hate.
A
You
know
we
got
a
lot
of
capacity,
it'd
be
really
nice
to
be
able
to
share
and
open
some
also,
some
really
cool
things
that
come
out
kelly
gaither's
been
doing
some
work
with
some
of
the
folks.
I
can't
remember
the
name
of
the
small
university,
but
it's
one
of
the
native
universities
in
hawaii
in
honolulu,
and
lots
of
really
really
really
great
stuff
coming
out
of
that
introducing
people
to
data
science
and
allowing
them
to
see
that
they
can
do
that
so
good
stuff
all
right.
A
D
Now,
I'm
here
at
you're
right,
oh,
you
are
okay,
sorry,
it's
shannon!
We
actually
canceled
the
last
meeting
for
so
you're,
not
missing
it
much.
You
know,
like
all
the
hubs
actively
involved
in
the
hdr
workshop
about
a
month
ago
and
following
up
on
the
report
out
that
also
now
that
that
is
over
and
renata's
hand
just
handed
the
leadership
of
the
ncc
over
to
john
in
the
midwest.
D
A
E
Yeah
we
do,
I
put
a,
I
put
a
link
in
the
chat
for
people
to
check
out.
If
you
haven't
seen
it
already,
we
are
getting
ready
to
host
a
a
one
day
meeting
and
the
details
of
that
are.
Are
there
it's?
The
topic
is
big
data
promises
and
obstacles,
agricultural
data,
ownership
and
privacy,
and
so
you
see
the
see
the
lineup
and
the
agenda
we're
working
on.
E
In
fact,
I
was
just
working
on
it
before
I
joined
this
meeting
with
the
with
the
other
organizers
and
so
organizing
committee
is
me,
cynthia
parr
from
usda
aaron
bergstrom
from
university
of
north
north
dakota,
and
also
now
the
digital
sport,
pi,
jason,
capela,
pepsico
and
phil
pardee
of
the
university
of
minnesota
is
a
he's,
an
agro-economist,
and
so
we've
done
a
few
things
on
this
topic.
If
you're
interested
in
issues
specifically
around
data
ownership
and
privacy
in
lag
space,
this
would
be
a
good.
A
A
F
F
F
Sorry,
that's
spam,
so
covet
information,
commons
and
we're
going
to
do
a
demo
today
and
I'm
going
to
give
you
all
a
link
to
the
demo
with
the
password
so
that
you
can
go
in
and
try
it
out
and
there's
a
feedback
form
on
it,
because
this
is
a
soft
launch
and
we
got
the
award
the
beginning
of
may.
This
is
our
soft
launch
and
we're
supposed
to
launch
the
mvp
the
week
of
july
10th
and
it
has
some
powerful
stuff
and
which
means
it's
kind
of
complex,
and
so
we
want.
F
F
Next
week
in
zoom
enabled
different
segments,
and
we
just
found
out
that
we're
going
to
have
a
bd
hubs,
pi
and
ed,
an
sf
see
fun
steering
committee
chair
meeting,
probably
one
day
in
the
september
time
frame,
so
john
mcmullen
just
reached
out
to
the
eds
to
ask
them
to
try
to
coordinate,
and
then
we
just
got
our
seed
fund
steering
committee
chair
onboarded
a
couple
weeks
ago,
and
that's
so.
We
can
figure
out
what
to
do
this
to
with
this
250
000
a
year
that
we
get
for
seed
fund
pilot
programs.
C
I
could
talk
just
briefly
about
the
eastern
regional
network.
All
hands
meeting
the
eastern
regional
network
is
is
a
coalition
of
the
willing
across
the
northeastern
states,
starting
in
delaware,
which
is
almost
northeast
up
through
maine,
which
is
solidly
northeast,
but
stops
short
of
north
nova
scotia,
and
the
idea
is:
how
do
you
lower
barriers
to
collaborative
research
across
campuses
within
the
region
and
the
barriers
that
we
seek
to
lower
are
people?
You
know
how
do
you
get
with
facilitators
to
work
across
campuses
compete
resources?
C
So
how
do
you
make
it
as
easy
to
get
to
a
resource
on
a
different
campus
as
it
is
to
get
to
an
eduroam
wireless
wireless
accounts
today
and
data?
How
do
you
you
seamlessly
share
data
when,
when
you
need
to
between
collaborators
or
groups
of
collaborators
across
campus,
we've
been
kind
of
beavering
away
at
this
for
a
couple
of
years?
This
is
our
second
annual
all
hands
our
first
virtual
all
hands
and
we'll
be
kind
of
covering
all
those
topics
in
kind
of
little
short
sessions
across
the
across
the
week.
A
B
B
We
also
had
some
proposals
under
review
from
partners
who
integrated
their
work
into
or
who
hoped
to
integrate
what
they're
doing
and
with
osm.
So
I
think
it's
been
a
really
positive
outgrowth
of
the
work
of
the
hubs
as
well
as
in
this
next
phase.
We
would
really
want
to
double
down
and
partner
more
with
the
hubs
to
do
outreach
to
find
projects
that
need
something
like
this,
especially
so
they
can
put.
C
A
Sounds
good,
which
I
think
carries
right
into
sort
of
the
next,
which
is
any
updates
on
sort
of
outreach
or
our
other
parts
that
we're
doing
around
data
here,
and
I
mean
I
think
that
may
be
something
we
need
to
focus
a
little
bit
towards
the
fall
once
getting,
osm
and
and
other.
I
think
you
know
the
minority
serving
universities
and
and
community
colleges
and
data
science
around.
A
F
Is
it
no?
No,
I
I'm
not
being
I'm
just
saying,
because
I
want
to
share
something
in
particular:
there's
a
young
man
younger
than
I
am
anyway
that's
getting
his
phd
in
neuroscience
at
princeton
who
reached
out
to
me,
and
he
actually
teaches
a
neuroscience
class
online
to
hispanic
speaking
students
in
his
spare
time.
While
he's
getting
his
phd
in
neuroscience.
F
Not
yeah
well
brilliant
people,
that's
what
they
do.
What
else
can
I
do?
I'm
so
bored
and
so
we're
working
with
the
new
york
hall
of
science
to
see
if
we
can
create
some
kind
of
program?
He
already
did
it
with,
I
think
harvard
and
some
in
some
groups
around
there.
So
maybe
there's
something
cool
we
can
do
with
that
speaks.
You
know
spanish
as
a
native
language.
F
So
that's
the
very
cool
idea
we
can
think
about,
and
then
the
other
thing
is
there's
this
carpentries
organization
that
some
of
the
hubs
have
been
partnering
with
I'm
brand
new
with
this.
But
one
of
the
things
we're
thinking
about
is
leveraging.
They
have
a
library
carpentry.
They
have
a
data
science
carpentry,
a
software
carpentry
in
the
library
carpentry
and
we're
thinking
of
maybe
trying
to
leverage
the
library
carpentry
to
get
to
underserved
communities,
and
so
we're
going
to
talk
with
our
new
york
hall
of
science
friends
there
in
corona
queens.
F
B
So,
instead
of
a
holy
librarian
audience,
people
that
are
ci
people
or
scientists
themselves,
postdocs
grad
students
and
chris
erdman
is
at
rentzy
and
he
headed
up
library,
carpentry,
and
so
we've
got
some
stuff
that
we're
trying
to
tee
up
for
this
coming
year
in
the
west,
tub
that'll
be
based
on
that.
We
should
definitely
work
together.
F
That
sounds
great.
Why
don't
I
send
you
an
email
and
we
can
have
a
call
anyone
else.
You
want
to
pull
in
because
katie
and
I
are
still
trying
to
figure
it
out.
I
actually
audited
a
train.
The
trainer
so
their
day
just
get
a
feel
for
what
they
do
and
it
was.
I
was
pleasantly
surprised
so
yeah.
Why
don't
I
send
you
an
email?
F
C
That
I'm
a
copy
I
with
with
carol
johnson,
who
now
you
may
have
gotten
to
to
know
she's
now
part
of
tech
on
something
called
esap
or
advancing
computer
science,
education,
they're
working
mainly
at
the
k
through
12
level.
It's
something
like
you
know,
40
of
the
of
the
50
states
and
are
working
largely
on
driving
policy
and
driving
computer
science,
education
and,
and
their
part
of
their
theme
is,
is
how
about
we
do
dei
at
the
beginning,
as
opposed
to
an
add-on.
F
F
A
A
You
know
the
right
data
to
the
right
data
scientists
to
the
right
communities
so
that
they
see
yes,
I
can
do
this
and
really
change
the
world
around
me,
which
is
important
right
now
in
many
ways.
So
yeah
I
will
let
me
I
can
talk
to
carol
and
other
folks
as
well.
If
I
see
them
in
person
ever
again,
it's
sort
of
the
problem
when
you've
got
185
people
all
on
zoom,
you
forget
half
of
them
because
you
don't
see
them
daily.
A
So
why
don't
we
go
ahead
and
get
started
with
that,
and
so
the
first
presentation
that
we
have
is
by
peter
rose
and
elias
might
know.
I
was
going
to
say
something
when
I
said
zodlowski
my
tongue
just
gets
tied
up
around
season.
L's,
I'm
not
I'm
not
yeah
anyway.
Sorry
about
that.
I
have
the
name
niall
gaffney,
I'm
used
to
it
being
mangled
as
well
so,
but
they
are
going
to
be
talking
about
the
structural
bioinformatics
lab
work
there.
A
A
Other
aspects
of
sort
of
drug
discovery
and
and
other
key
facets
of
what
we're
very
interested
in
right
now
and
elias
works
in
the
is
the
director
of
spatial
information
systems
at
the
university
of
california,
san
diego,
where
he
works
on
many
different
sort
of
spatial
information
pieces
of
all
of
this
and
they're
going
to
talk
to
us
about
cobit
19
net
and
what
they're
doing
right
now.
So
I
will
turn
it
over
to.
I
believe.
Peter's
talking
and
ilia
is
doing
the
slides.
G
All
right
is
that
correct
yeah
we
both
present
thanks
for
the
introduction.
First,
I
want
to
thank
christine
for
inviting
us
to
the
data
hubs
meeting.
It
sounds
really
interesting
what
you
guys
are
doing,
and
hopefully
maybe
we
find
some
connections.
You
know
along
the
way
as
we
present
here,
so
we
will
talk
about
our
effort
to
develop
a
heterogeneous
network
of
covate
19
data,
so
I'm
peter
rose
as
mentioned.
I
run
the
structure
by
informatics
lab,
so
I'm
on
the
on
the
bios
side
and
earlier
work
more
on
the
geographical
side
of
things.
G
So
this
was
a
good
project
where
we
can
really
merge
different
disciplines
together
and
the
way
this
happened
is
through
a
grant
that
we
have
from
the
nsf
called
the
nsf
convergence
accelerator.
So
we
just
finished
a
planning
grant
a
few
days
ago,
and
the
idea
of
this
open
knowledge
network
is
to
to
integrate
data
from
multiple
disciplines.
Together.
G
You
know
you
have
all
those
data
silos
and
you
know
how
can
we
connect
them
together,
so
the
nsf
funded
this
project
with
the
idea
is
that
each
project
team
needs
to
span
in
a
multiple
disciplines
and
multiple
organizations.
So,
for
example,
our
collaborations
is
with
ucsd
university
of
texas,
health
and
and
the
pangea
project
at
at
entire
as
an
example.
G
So
this
is
our
project
called
conquer,
and
here
we're
trying
to
bridge
three
disciplines
on
the
left-hand
side.
We
have
bio-medicine
and
health
data
and
we're
not
starting
from
scratch.
So
at
ucsd
we
have
worked
on
a
project
called
data
mat
which
is
a
discovery
engine
for
life
science.
Data
set
it's
basically
like
a
pubmed,
but
this
is
like
a
discovery
engine
for
life
science.
Data
sets
contains
about
2.3
million
data
sets
then,
on
the
other
hand,
earlier
worked
on
a
project
called
data
discovery
studio
where
basically
have
geoscience
data
set.
G
Also
in
about
1.7
million
data
sets
and
they're.
Basically,
all
kind
of
data
sets
that
have
a
geolocation
with
it
and
we
also
work
with
ncaa
which,
provided
you
know,
climate
data.
So
basically,
we
want
to
see
how
we
can
connect
all
those
you
know
different
data
sources
together,
so
you
can
run
a
query
across
multiple
disciplines.
So
that's
what
conquer
is
about
now?
The
concur
phase.
G
One
has
stopped
and
we
are
applying
now
for
conquer
phase
two,
but
in
the
meantime
we
got
a
rapid
grant
from
the
nsf
and
that
also
focuses
on
a
knowledge
network,
but
specifically
for
kovit
19,
where
we
can
integrate
multiple
heterogeneous
data
like
the
ones
shown
here.
So,
on
the
left
hand
side,
this
is
the
cathedral
in
san
paolo.
Here
we
have
a
cathedral
and
actually
the
other
way
around.
This
is
a
cathedral
in
milan.
This
is
a
cathedral
in
sao
paulo,
how
you
know
how
are
they
connected
they're,
probably
connected
to
some
air
travel?
G
Here
we
see
piece
of
a
genome
from
the
source
cove
to
virus.
We
see
some
repurposing
of
drugs
ppe's.
This
is
the
3d
protein
structure
of
the
spike
protein.
So
how
can
we
connect
this
all
together?
You
know
and
make
sense
out
of
that.
So
that's
what
we're
trying
to
do
so
we're
working,
not
alone.
So
there's
at
least
three
groups
involved
here,
so
in
blue.
This
is
us
ucsd,
so
in
any
of
those
spheres,
the
anything
that
has
a
blue
shade
in
it.
G
That's
something
we
involved
in
that
so
earlier
is
looking,
for
example,
at
population
characteristics
we're
looking
at
health
data,
I'm
working
more
on
the
biomedical
side,
so
I'm
looking
at
virus
strains
and
so
on
we're
looking
at
environmental
factors,
then
colleagues
from
uc
san
francisco,
they
are
looking
really
just
in
the
bio
area
like
genes
and
proteins
and
drugs
and
then
and
so
forth,
and
then
there's
another
rapid
grand
award
to
uc
santa
barbara
shown
in
green
here
and
they're,
focusing
more
on
supply
chain
and
transportation.
So
the
idea
is
those
three
networks.
G
G
So
for
our
part,
we
are
trying
to
link
basically
three
things
together.
One
we
have
the
host
which
in
our
case
the
human
being,
you
know
a
host
for
the
covet
disease.
Then
we
have
the
pathogen,
which
is
a
zoskov2
virus
and
everything
else
is
environment
in
which
the
host
and
the
pathogen
lives
and
some
of
the
data
we
want
to
integrate.
This
is
not
a
complete
list.
You
know
we
have
outcomes
like
confirmed
cases,
death
and
so
on.
We
have
patient
data
like
demographics,
not
for
specific
patients
or
for
patient
groups.
G
G
So,
as
I
said,
we
just
started
this
project
and
we
are
presenting
our
data
in
a
knowledge
graph
where
we
link,
you
know
different
nodes
together
and
right
now.
This
is
pretty
sparse.
We
don't
have
much
yet,
but
you
know
basically
what
this
shows.
On
the
left
hand
side.
Those
are
all
the
geo
locations,
starting
at
the
world
level.
We
break
down
the
world
into
un
regions
which
are
basic
continents
and
subcontinents
and
and
so
forth.
Then
we
have
countries.
The
countries
are
subdivided
in
admin,
one
zones
which
are
states
or
provinces.
G
Then
we
have
admin,
two
zones
which
are
counties
and
cities
and
and
so
forth.
So
basically,
we
have
mapped
the
entire
world
up
to
cities
up
to
a
thousand
citizens,
and
so
that
that's
basically
the
knowledge
graph.
That
represents
the
earth
and
we
will
go
even
deeper,
maybe
to
a
zip
code
level
and
census
tract
level,
but
we
are
not
there
yet.
Then,
on
the
right
hand,
side
I
show
a
little
bit
about
the
biology
of
covet
19.
So,
first
of
all,
we
have
cases
that
have
been
reported
here
in
different
areas.
G
You
see
reported
in
those
are
the
different
areas
where
we
we
find
case
of
this
outbreak.
This
outbreak
is
related
to
a
specific
organism
that
sascov2,
which
has
multiple
strains.
They
are
linked
to.
You
know
publication.
We
know
what
the
genome
is.
We
know
what
the
genes
are.
We
know
what
the
proteins
are.
The
strains
are
also
linked
to
specific
locations.
Those
you
know,
errors
are
missing
in
in
this
graph
here,
so
this
is
just
the
beginning.
G
So
the
idea
is,
we
will
integrate
more
data,
but
it
will
always
be
integrated
into
geo
locations
which
are
shown
on
the
left
here,
and
so
this
is
an
early
version
of
that
knowledge
graph
and
instead
of
showing
you
the
picture,
let
me
show
you
a
live
graph,
so
we,
our
graph,
is
represented
in
a
database
system
called
neo4j.
It's
a
data,
a
graph
based
data
management
system.
Here-
and
here
you
see
just
a
small
view
of
the
entire
graph
here.
So
what
we're
seeing
here?
G
First
of
all
over
here
this
node
in
in
this
light
blue
color.
This
is
human
being
that's
the
host
of
you
know
the
disease,
and
here
we
have
the
virus.
You
know
the
human
care
is
this
virus.
This
virus
has
many
strains
which
are
you
know
the
little
bubbles
around
here
that
are
moving
around
here.
Those
are
different
strains,
peter.
A
A
G
Yeah,
so
this
is
neo4j,
so
so
here
we
see,
this
is
the
source
virus
which
has
different
strains,
and
this
beige
color.
Here
we
have
the
human
being,
which
is
the
host,
and
then
the
genome
was
published.
This
is
for
the
first
case
in
in
wuhan,
where
they
actually
determined
the
genome,
and,
interestingly,
this
genome
came
already
out
in
in
january
of
this
year
and
the
data
for
that
we
already
collected.
Actually,
you
could
see
that
on
the
bottom
here
on
on
december
5th,
which
is
interesting,
so
they
knew
very
early
on.
G
You
know,
what's
what's
going
on
here,
so
this
represents
the
first
genome.
That
genome
has
various
genes
in
in
queen
here,
and
those
genes
are
expressed
into
proteins,
and
you
have
a
lot
of
heard
about
the
spike
protein.
That's
this
red
part
that
you
always
see
in
tb
on
around
the
virus.
That's
the
spike
protein
is
this
one
here
and
that
interacts
with
various
human
proteins
shown
on
the
right
hand,
side
one
of
them
is
angiotensin
ii,
which
is
this
note
right
here?
You
know
that's
basically
how
the
virus
gets
entry
in
into
the
human.
G
On
the
other
hand,
then,
for
each
strain
each
strain
has
various
mutations
they
are
listed
here,
and
this
is
a
specific
mutation
found
in
singapore.
So
we
you
see
how
we
link
together
information
about
location,
specific
virus
strains.
We
link
that
back
to
you,
know
the
genome
to
the
protein,
so
this
is
just
a
small
view
of
the
whole
grass,
so
it's
much
larger
so
right
now
it
contains
about
460
000
nodes
and
about
800
000
relationships.
So
again,
this
is
just
you
know,
beginning
of
a
larger
prototype,
and
I'm
not
sure.
C
G
Yes,
you
say
now:
okay,
now
my
slides
don't
advance,
okay,
so
here's
an
example
of
of
a
study.
We
did
with
this
knowledge
graph.
You
know
so
in
in
this
panel
d.
Those
are
where
each
node
in
in
this
beige
color,
that
is
a
person
actually
a
person
that
lives
in
sao
paulo,
but
they
visit
visited
italy.
They
actually
visited
milan
and
caught
the
virus
here,
specific
strain
a2a,
and
then
they
traveled
back
home
to
sao
paulo
and
then
that
led
to
a
community
spread
which
is
shown
in
level
in
panel
c.
G
Where
now
you
have
people
in
sao
paulo,
they
all
carry
this
to
a
a
strain.
So
that's
an
example
in
how
we
can
visualize
this
in
a
knowledge
network
and
similarly,
that
same
strain
is
from
san
francisco
and
also
in
in
san
diego.
So
it's
just
a
small
use
case.
You
how
you
might
you
know,
use
such
a
knowledge.
We
have
so.
G
G
F
G
G
As
you
know,
volunteers,
but
we
want
to
expand
that
to
the
community,
so
we
want
to
make
it
really
easy
for
everybody
to
contribute
data
and,
first
of
all,
the
way
we
create
our
knowledge
graph.
It's
all
based
on
open
access
data.
We
don't
want
one.
Somebody
sent
us
a
data
file.
Well,
we
don't
really
know
where
it
came
from.
Everything
is
based
on
publicly
available
data.
So
that's
that's
one
thing.
You
know
because
we're
creating
what
we
call
an
open
knowledge
network.
We
don't
want
to
have
any
proprietary
data.
G
We
also
don't
like
the
idea
of
hand,
manipulating
data
so
going
from
the
raw
data.
We
have
workflows
to
automatically
create
curated
data,
and
that
is
done
through
jupiter
notebooks.
Since
you
know
many
people
know
how
how
to
use
them
and
they're
great
for
data
cleanup.
So
we
can
automate
the
data
clean
up
through
jupiter
notebooks
and
they're
all
open
source
they're,
all
in
github.
So
it's
completely
a
reproducible,
workflow
and
those
jupiter
notebooks
stick
create
just
simple
csv
files.
G
They
represent
the
nodes
and
relationships
which
are
then
uploaded
into
the
knowledge
graph,
and
this
happens
every
night
at
midnight.
Pacific
time
we
know
we
run
this
workflow
every
day
get
new
data
because
many
of
the
data
sources
you
know,
are
updated
on
a
daily
basis.
So
we
we
rerun
that
workflow
every
night
and
create
you
know
the
updated
knowledge
from
that.
So
you
have
always
fresh
data.
G
For
example,
a
lot
of
data
iron
free
texts
that
causes
problems,
ambiguities,
for
example,
dealing
with
location
data
such
as
you
know,
orange
county
florida
versus
orange
county.
California,
often
scientists
are
not
specific
enough
to
you
know
to
define
where
the
data
are
coming
from,
and
I
don't
want
to
go
through
all
the
roadblocks
here,
but
there
are
many
data
gaps,
they're
missing
data
and,
and
so
forth.
G
You
know-
and
so
one
thing
early
on,
we
thought
about
is
interoperability
for
knowledge
graphs,
because
it
doesn't
help
if
we
create
a
knowledge
work,
but
that
is
again
just
another
silo.
So,
for
example,
for
all
our
biological
data,
we
use
what
is
called
curase
or
compact
uris.
That
means,
if
you
have
a
specific
identifier
for
a
gene
or
protein,
etc.
You
have
a
compact
id
which
consists
of
a
prefix
and
then
a
session
number.
G
Those
are
linked
to
persistent
urls,
which
eventually
can
be
resolved
to
you
know
a
specific
information
resource
and
the
way
it
works.
For
example,
if
we
have
a
protein
here
that
protein
yp
so
and
so
comes
from
ncbi
protein
and
using
identifiers.org,
if
you
plug
in
this
id
here,
it
will
resolve
to
the
specific
data
set.
So
basically
that
way,
we
we
exactly
define
where
all
the
data
come
from
and
this
leans
on
a
model
developed
by
lawrence
berkeley
labs.
They
call
it
this
bio-link
model
and
I
want
to
skip
over.
G
This
is
just
an
example
how
how
we
use
those
qrs.
You
know
to
make
it
interoperable,
and
so
I
showed
you
this
graphic
viewer,
which
is
called
the
neo4j
browser
where
you
can
graphically
browse
this
network,
which
is
nice,
but
it's
not
really
reproducible
but
fortunately
there's
a
plugin
for
jupiter
notebooks.
So
you
can
run
a
query.
Here's
an
example
of
a
query:
that's
using
what's
called
the
cipher
query
language.
You
can
run
that
in
a
jupiter
notebook,
for
example.
G
Here
we
list
all
the
different
proteins
on
the
right
hand,
side
you
see
the
the
virus
with
the
different
components,
and
here
we
list
all
the
different
components
of
this
virus
as
an
example
here.
So
it's
very
nicely
integrates
with
jupiter
notebooks,
and
now
we
have
you
know
a
reproducible
pipeline
to
create
the
knowledge
graph,
a
reproducible
way
of
of
analyzing
the
data.
G
A
A
Okay,
okay,
because
we
still
have
two
more
to
go.
So
if
we
could
just
hit
the
high
points,
I'm
sorry
to
to
pull
the
rug
out
on
you
ilia.
But
you
know
I'm
sure
there
are
people
that
are
going
to
have
to
go
at
the
top
of
the
hour.
So
I
want
to
make
sure
that
that
florence
and
christine
have
their
opportunities
as
well.
H
H
The
notebook,
so
you
can
see
that
we
we
get
the
zip
code
here
and
then
we
can
run
something
such
as
run,
some
queries
against
the
knowledge
graph
or
update
knowledge
graph
and
so
on.
So
if
we
do
that,
then
we'll
be
able
to
connect
to
the
knowledge
graph
that
peter
just
described
and
start
querying
that,
and
the
problem
is
that
here,
we're
talking
at
the
level
of
zip
codes
and
the
knowledge
graph
doesn't
yet
have
information
about
that.
H
So
we
have
information
from
johns
hopkins
and
similar
worldwide
kind
of
sources
where
san
diego
is
just
a
name
of
something.
Well,
if
you
look
at
its
at
san
diego
locally,
it's
just
one
of
18
municipalities
within
san
diego
county
and
has
fairly
well
defined
boundaries,
so
that
makes
it
difficult
to
to
connect.
H
This
is
one
of
the
dashboards
that
we
have.
We
have
another
set
of
dashboards
which
I'm
going
to
share
with
you
on
on
the
chat,
and
here
what
you
can
do,
you
can
look
at
multiple
factors
that
we
are
entering
into
the
knowledge
graph
simultaneously
and
figure
out
which
areas
are
problematic
on
multiple
on
multiple
indicators.
H
So
we
have
a
group
of
we've.
Actually
have
some
experience
with
schema.org
markup,
we
have
about
900
000
records
about
geoscience
data
sets
that
are
picked
up
by
google
and
when
you
can
search
when
you
search
google
data
set
search,
it's
very
likely
that
you
hit
data
discovery
studio
on
multiple
things,
so
that's
the
work
we've
been
doing
for
for
the
county
and
actually
one
issue
that
may
be
of
interest
to
to
to
many
of
you.
H
People
have
been
talking
about
fair
data
management
that
has
not
really
applied
yet
to
knowledge
graphs,
but
the
number
of
knowledge
graphs
is
proliferating,
and,
as
I
mentioned,
we
have
a
knowledge
graph
that
works
at
the
level
of
well.
That
goes
down
to
counties,
but
then
we
have
additional
knowledge
graph.
That
would
be
for
san
diego.
Specifically,
how
do
we
query
across
of
them,
given
that
some
identifiers
may
not
match
definitions
of
the
variables
would
be
not
identical
and
also
we
may
expect
multiple
knowledge
graphs
to
exist
at
this
at
this
level.
H
H
How
we
can
make
them
interoperable
so
that
from
a
dashboard
like
that,
we
can
query
across
different
knowledge
graphs
and
they
talk
to
each
other
and
produce
some
result
and
how
we
establish
common
vocabulary.
Again,
it
can
be
schema,
torque
based
or
it
could
be
something
else.
So
on
that
I'll
finish,
thanks.
F
F
F
A
Oh,
that's
all
all
great
stuff.
We
should
probably
circle
back
on
some
of
this
when
we
have
a
full
hour
on
on
parts
of
this,
so
look
forward
to
hearing
more
this
in
the
future
so
rather
than
rambling
on.
I
will
let
florence
go
ahead,
you're
next
on
the
schedule,
and
I
think
we
all
sort
of
know
who
you
are
so:
okay,
your
azaleas,
take
it.
F
F
F
Okay,
let
me
show
my
that's
really
weird.
Okay,
I've
never
had
this
happen
before
okay
live
in
there.
Can
you
see
this
now?
No
okay.
So
this
is
the
covet
information
commons,
and
this
is
based
on
a
rapid
that
we
got
at
the
northeast
big
data
hub
to
work
with
the
other
hubs
and
the
idea
is
to
create
an
open
resource
regarding
nsf
funded
research,
addressing
the
cobin
19
pandemics,
starting
with
the
rapid
awards.
F
So
the
idea
is
that
pis
would
use
it
and
sf
people
presidents
of
universities
who
want
to
see
how
they're
doing
versus
other
universities
on
getting
code
funding
and
things
like
that,
and
there
are
two
basic
ways
we
have
of
searching
at
this
point.
One
is
this
really
cool,
coveted
research
explorer
tool
which
is
actually
based
on
lingo
4g
explorer,
if
you're
familiar
with
that,
and
it
is
actually
an
internal
nsf
tool,
they've
exposed
to
us
to
use
on
this
covet
information.
Commons.
Can
you
see
this
map.
F
Yes,
okay,
great,
my
my
screen's
getting
funky
hang
on
a
second
okay,
so
this
map
actually
is
kind
of
a
topographical
map.
It's
machine
learning
based.
So
it's
looking
at
all
the
covid
rapid
awards
and
then
looking
at
what
categories
they
go
into,
then
you
could
look
at
what
they
call
a
tree
map
which
is
kind
of
like
the
periodic
table
of
the
elements.
You'll
see
what
I
mean
in
a
minute.
F
Very
interesting,
and
so
this
is
kind
of
by
thematic
areas,
and
then
you
can
also
see
a
list
of
the
awards
and
it's
not
the
easiest
thing
on
the
planet
to
use
so
we're
looking
for
input
on
the
user
interface.
So
as
an
example,
you
can
see
this
query
over
here.
Do
you
see
this?
You
know
fun
little
terminology
covert,
19
or
coordinating.
F
F
Know
well
so
they
had
to
actually
query
their
database
to
realize
they
weren't
getting
all
the
awards
because
they
have
to
use
these
other
words,
so
they
did
that.
But
then
let's
say
that
you
wanted
to
look
at.
You
know
these
types
of
things
and
you
wanted
to
add
end
use.
Boolean
algebra
mutation,
like
my
friend
at
princeton,
wants
to
do.
F
Then
you
put
that
in
and
then
you
analyze
again
and
then
it
comes
back
with
a
much
smaller
number
and
then
you
can
look
at
a
tree
map
of
it
and
you
can
actually
see
what
universities
are
working
in
these
different
areas.
So
this
is
one
of
the
tools
and
I'm
going
to
stop
sharing
for
a
minute,
because
my
my
laptop
is
misbehaving.
I
have
to
put
it
back
under
control.
Okay,
so
that's
one
of
the
ways
you
can
get
into
the
data
and
this
is
an
open
source
tool.
F
It
actually
goes
against
the
entire
nsf
database,
but
we're
not
telling
everybody
that,
but
they
can
know
that
because
it's
all
public
data
but
we're
starting,
you
know
we're
using
it
first
for
covid
and
coronavirus
longitudinally.
The
other
tool
that
you
can
use
is
we've
actually
just
done.
You
know
quick,
quick
searches
through
the
simple
search
mechanism.
So
if
you
want
to
look
for
coven
19
awards
in
the
biological
sciences
as
an
example,
this
will
bring
you
to
the
simple
search.
F
So
you
can,
you
know,
search
for
them
and
then
what
we're
looking
to
do
is
to
actually
create
another
view,
and
let
me
see,
I'm
probably
not
going
to
be
able
to
share
it
and
find
it.
But
another
view
that
gives
you
an
easier
way
to
look
at
all
the
information
about
an
award
and
it's
brought
to
you
by
the
four
big
data
hubs,
and
so
what
I'd
like
to
do
is
to
send
to
the
ci
list
the
actual
the
url
for
this
with
the
password
and
id.
So
you
guys
can
try
it
in.
F
Ladies
and
there's
a
feedback
mechanism
on
the
bottom
of
the
website,
we're
in
soft
launch
timing,
I'm
looking
for
input
by
june
20th
and
we're
supposed
to
launch
the
nbc
nvp
the
week
of
july
10th,
and
I
would
love
your
input
on
it.
The
other
thing
we're
going
to
be
doing
is
sending
the
pi's,
like
you
peter,
for
the
coveted
rapids
a
survey
to
ask
you
for
additional
information.
You
would
like
us
to
populate
through
this
website,
so
like
some
of
those
cool
demos,
I'm
thinking,
oh
my
eyes
were
like
twinkling.
F
While
you
guys
were
presenting
your
stuff,
it
would
be
so
cool
to
be
able
to
link
to
that
from
the
website.
So
we're
going
to
send
an
email
through
the
program
officers
to
all
the
rapid
covered
pis,
there's
like
over
500
of
them
now
and
we'll
see
what
we
get
back
and
then
see
how
to
organize
it
on
the
website.
But
we
would
love
to
say
you
know:
here's
their
project
website.
F
A
B
When,
if
I
could
ask
you
chris
to
go
ahead
and
share
the
slides
and
just
move
on
move
ahead
to
slide
two
and
I'll
just
get
talking,
thank
you
for
doing
that
again.
So
I've
been
involved
in
an
initiative.
I'm
really
excited
to
tell
you
about.
I
wish
you
know
we
didn't
have
this
cova
thing,
but
there
are
some
silver
linings.
I'd
already
been
involved
in
go
fare
for
a
while,
which
you
might
know
what
fair
stands
for:
findable,
accessible,
interoperable,
reusable
ellie
and
peter
mentioned
this
concept.
B
Gofare
is
an
initiative
that
where
the
ghost
stands
for
global
and
open
and
their
companion
organization,
that's
synergistic
with
codata
and
rda
and
and
I've
got
actually
a
graphic
at
the
end.
That
shows
you
shows
you
a
bit
about
where
they
lie.
One
of
the
hallmarks
of
gofare
is
it
they
create
implementation
that
works,
so
they
do
things
like
take
the
rda
guidelines
and
outputs
and
try
to
put
them
into
practice.
For
example,
they're
very
big
on
promoting
machine
actionable
data.
The
idea
that
this
is
a
common
denominator
for
fair
data.
B
If
your
computer
can
read
the
data,
then
people
probably
can
too
and
about
march
15th
the
virus
outbreak
data
network
was
established,
the
one
of
the
founders
of
gopher
baron
bonds.
His
background
is
in
malaria,
research
in
africa,
and
so
he
had
this
idea
that
we
shouldn't
just
be
reactive
to
covid.
This
should
be
something
that
you
know
can
survive
and
help
us
in
other
crises.
We
don't
want
kova
to
survive,
just
just
the
vote
in
the
next
slide.
Please.
B
So
a
lot
going
on
in
this
slide.
This
is
one
of
the
pilots
underway
by
vodan
a
lot
of
the
activity
you
see
in
a
lot
of
the
demos
you'll
see,
especially
in
the
early
weeks
of
the
crisis,
we're
all
about
the
bottom
left.
You
know,
how
can
we
parse
journals
right?
Everyone
was
working
on
that,
but
we're
trying
to
remember
that
the
the
eventual
goal
is
to
get
up
to
this
top
right.
B
That's
out
there,
that's
related
to
covet
19
and
related
pandemics,
but
also
incorporates
that,
with
the
fair
data
points
which
I'm
going
to
talk
about
in
a
minute
puts
that
in
a
place
where
it
can
be
annotated
and
then
forwarded
on
in
a
much
better
view
to
be
used
by
analysis
platforms.
In
this
case,
there's
one
called
uritos,
which
is
a
spin-off
of
some
other
earlier
efforts
by
the
same
with
the
same
partners.
So
next
slide.
Please.
B
So
vote
on
africa.
Voter
in
africa
is
an
offshoot
of
the
the
bodan
implementation
network.
A
lot
of
us.
You
know
we
don't
have
people
that
consult
in
our
families.
Who've
lived
through
a
pandemic.
This
is
very
new
to
us,
but
for
africans
they
lived
through
ebola
not
too
long
ago,
and
so
they
have.
They
are
well
aware
that
this
is
something
that
they
need
to
think
about
and
use
for
whatever
the
next
health
crisis
is.
B
B
It's
for
africa
by
african
partners
and
it's
a
lot
of
people
working
at
country
levels,
health
ministries,
universities,
foundational
support
from
the
phillips
foundation
and
with
sponsorship
from
different
governments,
the
three
projects,
and
especially
as
relate
to,
I
think
our
interests,
people
on
the
call
they're
the
first
to
do
widespread
metadata
for
machines
workshop.
This
idea
that
fair
data
is
machine.
Actionable
data
they've
also
launched
to
train
the
trainers
program
for
creating
a
data
steward
workforce.
B
They
found
that
there
aren't
enough
people,
just
as
we
don't
have
in
the
u.s
or
the
eu
or
australia
and
new
zealand
don't
have
enough
data
stewards.
They
definitely
don't
either
and
then
also
setting
up
fair
data
points
which
are
places
that
algorithms
can
query
for
metadata
and
summary
data
and
the
idea
being
that
these
are
infrastructure
left
behind
for
future
crises.
B
B
So
this
is
just
a
quick
look
at
the
participants:
tunisia
and
the
north
nigerian,
the
west,
uganda,
kenya,
ethiopia
in
the
east
and
then
zimbabwe
in
the
south,
and
they
definitely
have
commonalities
in
their
challenges,
but
also
some
very
different
things,
especially
if
you're
looking
at
tunisia,
which
is
right
there
on
the
border
with
libya.
Next
slide,
please,
and
so
we
had
a
kobit
19
and
fair
data
webinar
that
peter
and
ilya
were
part
of
and
the
vote
in
africa.
B
B
Some
countries,
liberia
who's,
not
mentioned
as
an
official
partner,
but
has
partners.
There
is
also
looking
to
track
alcohol
abuse
and
domestic
violence
alongside
the
covenant
infections,
because
they're
seeing
some
correlations
and
they
urged
everyone
as
they're
thinking
about
these
data
infrastructures,
we're
building.
You
know
what
is
going
to
come
after
cobit
and
let's
make
sure
that
we
have
the
data
needed
to
respond
to
that.
B
Kenya
raised,
you
know
what
it's
like
to
try
and
manage
an
outbreak
in
a
slum
where
people
aren't
able
to
wash
their
hands
and
then
another
another
thing.
That
was
a
commonality
that
we
take
for
granted.
That,
I
believe,
is
on
the
some
people
put
on
the
list
of
human
rights,
along
with
with
water
and
electricity
having
unique
ids,
it's
very
hard
to
have
electronic
health
records
with
calpos,
but
on
the
positive
side
we
had
partners.
B
B
B
B
If
you
want
to
still
do
that
and
then
lastly,
to
set
up
something
like
vote
in
africa
is
as
simple
and
as
hard
as
finding
key
individuals
who
are
passionate
about
doing
that
and
all
of
the
projects
we
talked
about,
I
think,
are
the
same
way
and
will
be
the
same
way
years
from
now.
Success
depends
on
individuals
so,
with
the
last
slide,
there's
a
couple
links
if
you
want
to
find
out
more.
If
you
are
interested
in
this
topic,
I
can't
recommend
enough
joining
the
midas
network.
B
A
A
There's
just
that
all
of
this
stuff
is
is
interrelated
and
can
be
applied
in
in
places
all
around
and
very
important
for
that.
So
so
thank
you,
everybody,
and
if
we
want
to
have
sort
of
a
little
question
and
answer
discussions
here,
feel
free.
If
not,
we
can
stop
playing
hollywood
squares
for
a
little
while
and
get
started
on
the
weekend.