1. New
Approaches
for
iden1fica1on
and
selec1on
of
therapeu1c
targets
for
Complex
Disease
Stephen
H
Friend
MD
PhD
Sage
Bionetworks
Alzheimer’s
Disease
Research
Summit
May
14-‐15
2012
NIH
2. Disease
Preven1on
and
Treatment
• To
Prevent
need
to:
– Have
clinical
&
molecular
defini1on
of
disease
– Be
able
to
predict
progression
– Have
drugs
that
target
mechanisms
that
drive
progression
• To
Treat
need
to:
– Have
clinical
&
molecular
defini1on
of
disease
– Disease
modifying
therapies
For
Alzheimer’s
we
need
work
to
develop
all
of
these!
3.
4. Data-‐driven
Target
Iden0fica0on
If
we
accept
that
disease
is
driven
by
the
complex
interplay
of
gene1cs
and
environment
mediated
through
molecular
networks…….
Gene1cs
Gene1cs
Disease
progression
Disease
Modifying
Therapy
Healthy
Disease
Environment
Environment
State
State
………………………….then
it
follows
we
must
study
these
networks
and
how
they
respond
to
perturbagens,
how
they
differ
in
disease,
etc
5. Data-‐driven
Target
Iden0fica0on
If
we
accept
that
disease
is
driven
by
the
complex
interplay
of
gene1cs
and
environment
mediated
through
molecular
networks…….
Gene1cs
Gene1cs
Disease
progression
Disease
Modifying
Therapy
Healthy
Disease
Environment
Environment
State
State
………………………….then
it
follows
we
must
study
these
networks
and
how
they
respond
to
perturbagens,
how
they
differ
in
disease,
etc
6. Problem
is
Complex
and
will
not
be
solved
by
any
one
group
– New
Capabili1es
• Informa1on
Commons
• Portable
Legal
Consent
– New
Ways
to
Work
Together
• Public-‐Private
Partnerships
eg
ADNI
– Recognize
new
Roles
for:
• Pa1ents
• Ci1zens
• Funders
• Scien1sts
7. Two
recurring
problems
in
AD
research
Ambiguous
pathology
Diverse
mechanisms
Are
disease-‐associated
molecular
systems
&
How
do
diverse
muta1ons
and
environmental
factors
genes
destruc1ve,
adap1ve,
or
both?
combine
into
a
core
pathology?
Boom
line:
We
need
to
iden1fy
causal
factors
Boom
line:
There
is
no
rigorous
/
consistent
global
vs
correla1ve
or
adap1ve
features
of
disease.
framework
that
integrates
diverse
disease
factors.
7
8. Two
recurring
problems
in
AD
research
Ambiguous
pathology
Diverse
mechanisms
Are
disease-‐associated
molecular
systems
&
How
do
diverse
muta1ons
and
environmental
factors
genes
destruc1ve,
adap1ve,
or
both?
combine
into
a
core
pathology?
Boom
line:
We
need
to
iden1fy
causal
factors
Boom
line:
There
is
no
rigorous
/
consistent
global
vs
correla1ve
or
adap1ve
features
of
disease.
framework
that
integrates
diverse
disease
factors.
One
consequence…
"There
are
very
few
new
molecular
en22es,
very
few
novel
ideas,
and
almost
nothing
that
gives
any
hope
for
a
transforma2on
in
the
treatment
of
mental
illness.”
-‐
Thomas
Insel,
Science
2010
8
9. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
-‐
correlated
expression
of
mul1ple
genes
across
many
pa1ents
-‐
coexpression
calculated
separate
for
Disease/healthy
groups
-‐
these
gene
groups
are
ofen
coherent
cellular
subsystems,
enriched
in
one
or
more
GO
func1ons
Data
source:
Harvard
Brain
Tissue
Resource
Center
SNPs,
Gene
Expression,
Clinical
Traits
AD
n
=
284
Pre
Frontal
Cortex
Control
153
AD
168
Visual
Cortex
Control
116
AD
220
Cerebellum
Control
122
10. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
-‐
correlated
expression
of
mul1ple
genes
across
many
pa1ents
-‐
coexpression
calculated
separate
for
Disease/healthy
groups
-‐
these
gene
groups
are
ofen
coherent
cellular
subsystems,
enriched
in
one
or
more
GO
func1ons
Alzheimer’s-‐specific
regulatory
rela1onship
Microarray
result
Transcription
factor
Gene A Gene B
11. Where
does
coexpression
come
from?
What
does
a
“link”
in
these
networks
mean?
• What
is
the
evidence
that
coexpression
is
produced
by
regulatory
rela2onships?
• Gene
coexpression
has
mul1ple
biophysical
sources:
1:
Transcrip1onal
overrun
/
chromosome
loca1on
(Ebisuya
2008)
2:
Common
transcrip1on
factor
binding
sites
(Marco
2009)
3:
Epigene1c
regula1on
(Chen
2005)
4:
3D
Chromosome
configura1on
(Deng
2010)
Chromosome
segment
– Varia1on
in
cell-‐type
density
(Oldham
2008)
#1
#4
#2/TF
Gene
A
Gene
B
Gene
C
Promoter
x
Promoter
y
#3
11
12. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
-‐
correlated
expression
of
mul1ple
genes
across
many
pa1ents
-‐
coexpression
calculated
separate
for
Disease/healthy
groups
-‐
these
gene
groups
are
ofen
coherent
cellular
subsystems,
enriched
in
one
or
more
GO
func1ons
Example
“modules”
of
coexpressed
genes,
color-‐coded
13. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
2.)
Priori1ze
the
disease-‐relevance
of
the
modules
by
clinical
and
network
measures
Priori1ze
modules
through
expression
synchrony
with
clinical
measures
or
tendency
too
reconfigure
themselves
in
disease
vs
14. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
2.)
Priori1ze
the
disease-‐relevance
of
the
modules
by
clinical
and
network
measures
Priori1ze
modules
through
expression
Combina1on
of
cogni1ve
func1on,
Braak
score,
synchrony
with
clinical
measures
or
tendency
cor1cal
atrophy
with
differen1al
expression
too
reconfigure
themselves
in
disease
and
differen1al
coexpression
rank
modules.
vs
15. Iden1fying
key
disease
systems
and
genes
1.)
Iden1fy
groups
of
genes
that
move
together
–
coexpressed
“modules”
2.)
Priori1ze
the
disease-‐relevance
of
the
modules
by
clinical
and
network
measures
3.)
Incorporate
gene1c
informa1on
to
find
directed
rela1onships
between
genes
Priori1ze
modules
through
expression
Infer
directed/causal
rela1onships
synchrony
with
clinical
measures
or
tendency
and
clear
hierarchical
structure
by
too
reconfigure
themselves
in
disease
incorpora1ng
eSNP
informa1on
(no
hair-‐balls
here)
vs
16. Example
network
finding:
microglia
ac1va1on
in
AD
Module
selec0on
–
what
iden0fies
these
modules
as
relevant
to
Alzheimer’s
disease?
The
eigengene
of
a
module
of
~400
probes
correlates
with
Braak
score,
age,
cogni1ve
disease
severity
and
cor1cal
atrophy.
Members
of
this
module
are
on
average
differen1ally
expressed
(both
up-‐
and
down-‐regulated).
Evidence
these
modules
are
related
to
microglia
func0on
The
members
of
this
module
are
enriched
with
GO
categories
(p<.001)
such
as
“response
to
bio1c
s1mulus”
that
are
indica1ve
of
immunologic
func1on
for
this
module.
The
microglia
markers
CD68
and
CD11b/ITGAM
are
contained
in
the
module
(this
is
rare
–
even
when
a
module
appears
to
represent
a
specific
cell-‐type,
the
histological
markers
may
be
lacking).
Numerous
key
drivers
(SYK,
TREM2,
DAP12,
FC1R,
TLR2)
are
important
elements
of
microglia
signaling.
Alzgene
hits
found
in
co-‐regulated
microglia
module:
17. Figure
key:
Five
main
immunologic
families
found
in
Alzheimer’s-‐associated
module
Square
nodes
in
surrounding
network
denote
literature-‐supported
nodes.
Node
size
is
propor2onal
to
connec2vity
in
the
full
module.
Core
family
members
are
shaded.
(Interior
circle)
Width
of
connec2ons
between
5
immune
families
are
linearly
scaled
to
the
number
of
inter-‐family
connec2ons.
Labeled
nodes
are
either
highly
connected
in
the
original
network,
implicated
by
at
least
2
papers
as
associated
with
Alzheimer’s
disease,
or
core
members
of
one
of
the
5
immune
families.
22. Current
AD
projects
with
Sage
in
collabora1on
Follow-‐up
microglia
experiments
Confirming
TYROBP
relevance
in
human-‐derived
microglia-‐neuron
co-‐culture
Similar
microglia
experiments
with
Fc
receptor
(Neumann,
Gaiteri)
Novel
genes
validated
with
in
vitro
and
in
vivo
model
systems
Cell
culture
&
transgenic
FAD
crosses
with
novel
gene
KO’s
(Wang,
Kitazawa,
Gaiteri)
Addi0onal
microarrays
from
model
systems
Check
network
predic2ons
to
refine
both
algorithm
&
biology
(Schadt/Neumann)
Larger
cohorts,
proteomics
Building
networks
in
3x
larger
dataset,
newer
plaorm,
w/
detailed
clinical
info
(Myers,
Gaiteri)
23. Design-‐stage
AD
projects
at
Sage
Fusing
our
exper1se
in…
Gene
regulatory
networks
Diffusion
Spectrum
Imaging
Feedback
Microcircuits
&
neuronal
diversity
To
build
mul1-‐scale
biophysical
disease
models.
Join
us
in
uni1ng
genes,
circuits
and
regions!
Contact
chris.gaiteri@sagebase.org
24. List of 50 Influential Papers in Network Modeling
http://sagebase.org/research/resources.php
25. Now add Dimensions of Circuits, Brain Regions, Individual Dynamic Heterogeneity,
And Longitudinal Variations
26.
27. Ul1mately
these
efforts
will
fail
without
more
ambi1ous
thinking
– Ac1vate
Pa1ents
• Pa1ents
want
to
be
involved,
to
fund
research,
to
direct
the
research
ques1ons,
to
hold
the
scien1fic
community
to
account
• Portable
Legal
Consent
– Collect
Large
Scale
Longitudinal
Data
• We
need
to
collect
the
right
kind
of
data.
Molecular
and
Phenotypic
in
a
longitudinal
fashion
on
10s-‐100,000s
of
individuals
• Real
Names
Discovery
Project
– Build
an
Informa1on
Commons
• Synapse
– Engage
in
Collabora1ve
Challenges
• Breast
Cancer
Challenge-‐
IBM/Google/
Science
Transl
Med
28. Why not share clinical /genomic data and model building in the ways
currently used by the software industry
(power of tracking workflows and versioning
29. sage bionetworks synapse project
Watch What I Do, Not What I Say Reduce, Reuse, Recycle
My Other Computer is Amazon
Most of the People You Need to Work
with Don’t Work with You
30.
31.
32. We
pursue
Alzheimer’s
Care
is
if
it
were
an
“Infinite
Game”
and
We
pursue
Alzheimer’s
Research
as
if
it
were
a
“Finite
Game”
33. We
pursue
Alzheimer’s
Care
is
if
it
were
an
“Infinite
Game”
and
We
pursue
Alzheimer’s
Research
as
if
it
were
a
“Finite
Game”
YET
We
should
pursue
Alzheimer’s
Care
is
if
it
were
a
“Finite
Game”
and
We
should
pursue
Alzheimer’s
Research
as
if
it
were
an
“Infinite
Game”
34. Who will build the datasets/ models capable of providing powerful
insights enabling disease modifying therapies?
Power
of
Collabora1ve
Challenges
Evolving
Models
from
Deep
Data
Driven
Longitudinal
Cohorts
in
Worldwide
Open
Informa1on
Commons
Ins1tutes
Industry
Founda1ons
NETWORK
PLATFORM
PPP
Or
??????
Scientists Physicians Citizens “Knowledge Expert”