Inferenc nce and dynamical modeling of regul ulatory networks - - PowerPoint PPT Presentation

Inferenc regul

The Secon L

nce and dynamical modeling of ulatory networks controlling hematopoiesis

JOSÉ TELES cond q-bio Summer School and Conference LANL – 23 July 2008

OUTL

• I. INTRODUCTION
• II. (BROAD) PROJECT DE
• III. CURRENT WORK
• IV. IMMEDIATE PROSPEC

TLINE

ESCRIPTION ECTS

PhD @ @ Lund

CARSTEN PETERSON

PhD @ @ Lund

Hemato topoiesis

Metcalf D. Blood Lines (2006)

Hematopoiesis: g general principles

Metcalf D. Blood Lines (2006)

Cell fate d

?

decisions

SELF-RENEWAL DIFFERENTIATION APOPTOSIS QUIESCENCE

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS”

uestions

F GENE EXPRESSION REGULATION

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS” DIFFERENTIATION

uestions

F GENE EXPRESSION REGULATION

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS” IRREVERSIBILITY DIFFERENTIATION

uestions

F GENE EXPRESSION REGULATION

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS” IRREVERSIBILITY BRANCHING DIFFERENTIATION

uestions

F GENE EXPRESSION REGULATION

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS” LINEAGE SWITCHING IRREVERSIBILITY BRANCHING DIFFERENTIATION

uestions

F GENE EXPRESSION REGULATION

Main qu

ORGANIZATIONAL PRINCIPLES OF

“STEMNESS” LINEAGE SWITCHING IRREVERSIBILITY BRANCHING DIFFERENTIATION CELL EXPANSION

uestions

F GENE EXPRESSION REGULATION

Gene expression n in hematopoiesis

Gene expression

• p

r

• g

s

n in hematopoiesis

MASTER REGULATORS

• Cell

lineages representing discrete “genetic programs” mutually exclusive and intrinsically robust.

• Transcription factors act in intricate circuits of

gene regulation, specifying the stable lineage- specific transcriptome.

Gene expression n in hematopoiesis

Gene expression n in hematopoiesis

Gene expression n in hematopoiesis

Gene expression signatures

Experimen

EPO, haemin, IL-3 GCSF, GMCSF, IL-3 low Neutrophil lineage Erythroid lineage FDCPmix cells (murine)

ental setup

Tariq Enver 3 low

Experimen

24-36 30 time points (7 days)

ental setup

6 hrs Tariq Enver

Experimen

30 time points (7 days) a) Mi

• b) Ch
• c) Ce
• ental setup

Microarray data:

• (N): 3600 diff exp genes (265 TFs)
• (E): 4500 diff exp genes (354 TFs)

ChIP-on-chip data:

• Key TFs (Gata1, Gata2, Pu.1, Fog-1)

Cell population counts (7 time points):

• Blasts (progenitor)
• Immature erythroid
• Immature neutrophils
• Macrophages
• Megakaryocytes

Transcriptional n

a) Mi

• b) Ch
• c) Ce
• network inference

Microarray data:

• (N): 3600 diff exp genes (265 TFs)
• (E): 4500 diff exp genes (354 TFs)

ChIP-on-chip data:

• Key TFs (Gata1, Gata2, Pu.1, Fog-1)

Cell population counts (7 time points):

• Blasts (progenitor)
• Immature erythroid
• Immature neutrophils
• Macrophages
• Megakaryocytes

Transcriptional n

a) Mi

• b) Ch
• c) Ce
• Probabilistic graphical models

(Friedman N, 2004)

Dynamic Bayesian Networks

(Perrin BE et al, 2003)

Probabilistic Boolean Networks

(Shmulevich I et al, 2002)

Mutual Information Based

(Basso et al, 2005)

Microarray + ChIP analysis

(Ernst et al, 2007)

network inference

Microarray data:

• (N): 3600 diff exp genes (265 TFs)
• (E): 4500 diff exp genes (354 TFs)

ChIP-on-chip data:

• Key TFs (Gata1, Gata2, Pu.1, Fog-1)

Cell population counts (7 time points):

• Blasts (progenitor)
• Immature erythroid
• Immature neutrophils
• Macrophages
• Megakaryocytes

Structura ral motifs

Swiers et al. Developmental Biology (2006)

Dynamical features

• Stable atractors of gene circuits represent states
• The discrete transitions in bistable biochem

differentiation (Laslo et al, 2006)

• Control by external regulatory signals (Enver et al,

a) Stochastic (“selective”) cell fate control: ce by the cell in a chance fashion. External signa b) Deterministic (“instructive”) cell fate con activating/repressing sets of genes via signal

Hematopoiesis as a case-study for gen

es of hematopoiesis

• f differentiation. (Cinquin & Demongeot, 2005)

mical systems underlie cell fate decision or

1998):

cell fates constitute preexisting programs adopted als act as survival/growth factors. ntrol: external signals impose the program by transduction cascades.

enetic control of lineage specification

Dynamical

Loose et al. Curr Opin Hematol (2007)

al modeling

Dynamical modeling: th

blast

GATA1 PU1 GATA1 PU1

the PU.1 / Gata1 switch

Erythroid benz+ Neutrophil

Dynamical modeling: th the PU.1 / Gata1 switch

GATA1 PU.1

Huang et al. Developmental Biology (2007)

Dynamical modeling: th the PU.1 / Gata1 switch

GATA1 PU.1

Huang et al. Developmental Biology (2007)

Dynamical modeling: th the PU.1 / Gata1 switch

GATA1 PU.1

Huang et al. Developmental Biology (2007)

Dynamical modeling: th the PU.1 / Gata1 switch

Roeder et al. J Theor Biol (2006)

Dynamical modeling: extend

GATA1 PU.1

nding the PU.1 / Gata1 switch

PU.1 Gata1 X Y

Curren

a) In silico microdissection of microarray

nt work

y data

Curren

a) In silico microdissection of microarray

FROM:

Raw expression data Cell type fraction measurements

Expre

nt work

y data

ression level Time Cell type 1 Cell type 2 Cell type 3

TO:

GENE X

Curren

a) In silico microdissection of microarray

nt work

y data

Curren

a) In silico microdissection of microarray

OPTIMIZATION PROBLEM: Standard least sq

• ne of the cell type-speci

Lyl1 – Erythro

nt work

y data

squares solution as the linear estimate for each cific gene expression levels

roid lineage

Curren

b) Cell population models

B IE

αB 1 - αB

nt work

IE E

αIE 1 - αIE

Future

a) In silico microdissection of b) Cell population models c) Dynamical modeling (Pu.1 / d) Network inference

re work

• f microarray data

/ Gata1 ?)

Acknowled ledgements