Breast cancer is a complex and heterogeneous disease Tumor samples Protein expression Clinical features Mutational status Adapted from TCGA, Nature 2012.

HPN-DREAM Breast Cancer Network Inference Challenge

Breast cancer is a complex and heterogeneous disease

Tumor samples

Protein expression

Clinical features

Mutational status

Adapted from TCGA, Nature 2012

Transcriptional Subtype

Breast cancer is a complex and heterogeneous disease

Tumor samples

Protein expression

Clinical features

Mutational status

Adapted from TCGA, Nature 2012

Transcriptional Subtype

• Genomic and epigenomic aberrations (mutations, copy number changes, etc) influence cancer development

• Collection of aberrations in an individual sample create a unique “biological context” that influences cell signaling

• Improved understanding of network function will lead to the development of more effective therapies

HPN-DREAM Challenge: How are signaling pathways deregulated across breast cancers?

Patient

Tumor

Cell Line

High-throughput screen of protein signaling dynamics

InhibitorsDMSO

FGFR1/3iAKTi

AKTi+MEKi..

Inhibitor N

Stimuli

Serum

PBSEGF

Insulin

FGF1HGF

NRG1IGF1

……

~200

Pro

tein

s

8 Stimuli 5

Treatments

MCF7…

……

……

~200

Pro

tein

s

8 Stimuli 5

Treatments

UACC812

……

…

……

~200

Pro

tein

s

8 Stimuli 5

Treatments

BT20

……

…

…~45 P

rote

ins 8

Stimuli

N Inhibitors

BT549

……

……

Time

4 C

ell

Lines

Inhibitor Stimulus

…0 5 1

54

7 Timepoints

Data generated by Reverse Phase Protein Array (RPPA)

4 cell lines x 8 stimuli = 32 biological contexts for network prediction

Hold out a subset of inhibitor data for assessment of network inference and timecourse predictions

Training Data (4) treatments)

Test Data (N-4) treatments)

FGFR1/3iAKTi

AKTi+MEKiDMSOAll Data (N

treatments)

Test1Test2

….TestN-4

Creating a “Gold Standard” for assessment of predictions

4545

Mimics key aspects and characteristics of the experimental dataGenerated from a dynamical signaling network modelInferred networks can be assessed against against a true gold standard with known network structure

Companion in silico challenge

Subchallenge 1: Network Inference

Task: Create a network where nodes represent phosphoproteins and directed edges represent causal relationships between the nodesAssessment: Score against held-out test data

Predict

Training Data

1A Experimental data: predict 32 context-specific networks1B In silico data: predict 1 networkComplete submission requires both A and B parts

Using inhibitor data to infer network structure

Causal edges: 1. predict that perturbing (ie, inhibiting) parent node A will induce change in child node B

Time

Node B

ab

undance

With A inhibitor

A B

Cell Line 1, Stimulus 1

Time

A B

Cell Line 2,

Stimulus 1

Node B

ab

undance

2. are context-specific, and vary with cell line and stimulus

Control

Subchallenge 2: Timecourse prediction

Training Data

Predict

Task: Build a dynamical model to predict phospho-protein trajectories following inhibition of test nodesAssessment: Score against held-out test data

TimePro

tein

A

bun

dance

2A Experimental data2B In silico data

Subchallenge 3: Visualization

Task: Devise novel approaches to represent high-dimensional timecourse datasetsAssessment: Crowd-based peer-review

Submit

Training Data

HPN-DREAM Challenge: Participation

237+ registered participants

Complete final submissions: SC1 Network Inference: 59 SC2 Timecourse Pred: 10 SC3 Visualization: 14

Collaborative Bonus Round to foster exchange of ideas and development of hybrid models

Some details of assessment and test data will not be released until after the close of the collaborative round

HPN-DREAM ContributorsAnalysis and scoring

Steven HillThomas Cokelaer*Sach Mukherjee

In silico data generationMichael Unger*Heinz Koeppl

Experimental data generation

Nicole NesserKatie Johnson-Camacho

Gordon MillsJoe Gray

*Paul Spellman

Challenge organizersLaura Heiser

Julio Saez-RodriguezThea Norman

*Gustavo Stolovitzky

Synapse developmentJay HodgsonBruce HoffMike Kellen

*Steven Friend

Heritage Provider Network

Jonathan Gluck

Poster: DREAM03synapse.org/#!Challenges:DREAM8

Seru m

PB

SEG

FIn

sul

inFG

F 1H

GF

NR

G 1IG

F1

Seru m

PB

SEG

FIn

sul

inFG

F 1H

GF

NR

G 1IG

F1

Seru m

PB

SEG

FIn

sul

inFG

F 1H

GF

NR

G 1IG

F1

Seru m

PB

SEG

FIn

sul

inFG

F 1H

GF

NR

G 1IG

F1

Seru m

PB

SEG

FIn

sul

inFG

F 1H

GF

NR

G 1IG

F1

DMSO Inhib 1 Inhib 2 Inhib 3 Inhib 4

Sustained responseTransient response

Pro

tein

sAn information rich timecourse