Oct 12, 2007 Research Review Day 1 Josh Stuart, Ph.D. Biomolecular Engineering UCSC Research Review Day 2007 Biological Discovery From Genetic Network.

Oct 12, 2007Research Review Day

1

Josh Stuart, Ph.D.Biomolecular Engineering

UCSC Research Review Day 2007

Biological Discovery FromBiological Discovery FromGenetic Network PerturbationsGenetic Network Perturbations

Reverse-EngineeringReverse-Engineeringby Knocking Downby Knocking Down


2

“software” of life

http://images.google.com/imgres?imgurl=cuhwww.upr.clu.edu/~esuarez/adem/harddisk.jpg&imgrefurl=http://cuhwww.upr.clu.edu/~esuarez/adem/tema3.htm&h=150&w=150&prev=/images%3Fq%3Dharddisk%26svnum%3D10%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8


3

Genomes to function

?Genome

Hair

Neuron


4

Genomes to function

Genome

Neuron

Hair

Gene switched “on”

Transcriptome Interactome

Genes signaling


5

Function fromGenetic Knock-downs

• Genome sequence provides complete parts lists• Allows targeting of specific genes

– Cloning

– RNA interference

• High-throughput technologies allow monitoring genome-wide responses to knock-down

• Phenotype gives clues about gene function


6

What’s a knock-down?


7

Genetic Knock-DownsRNA interference

RNAiA. Fire et. al 1998


8

Two Examples

• Infer disease pathways

• Predict genetic interactions


9

Reverse engineering by knocking-down




10Knock-downs to Understand Cancer Invasiveness

Lee Lab, GWUMC


11Knock-downs to Understand Cancer Invasiveness

1. Identify knock-downs that reverse cancer invasion

2. Genome-wide expression under knock-downs

3. Infer invasiveness network

4. Predict new genes involved in process

Go to step 1.


12

up-regulatedin knock-down

down-regulatedin knock-down

Sensitive, genome-wide phenotypes: DNA Microarrays

knock-down

normal cells


13

DRYWET

Iterative invasiveness network strategy


14

Infer Networkfrom Secondary Effects

Single Phenotype

Network GenesPerturbed by

RNAi or Knockout

Effect GenesMeasured by microarray


15

Expression under

ee ee

ee eeAA

BB ∆A

∆Bee

Expression under


16

Predictions for Colon Cancer Invasiveness

• Identified a putative signaling network• Expanded the list of candidates in the network• Testing candidates for loss-of-invasiveness

phenotype

conserved rolein cell-migration


17V. cholera Networks

Yildiz Lab, UCSC


18

New Biofilm genes


19

Two Examples




20

Function from catastrophe• Most genes are nonessential• Genes knocked down together give phenotype• Can we infer function from knock-down combos?


21

Gene Network Discovery

• build networks from all interactions

• discover function from a gene’s links

• understand bigger picture of gene regulation


22Understanding Gene Function through Modifier ScreensUnderstanding Gene Function through Modifier Screens

Wild Type

A

B

C

X

A

B / R

C

XWild Type

A

B

C

X

D

E

F

Wild TypePhenotype Phenotype

Roy Lab, Univ Toronto

B --- RWithin Pathway Link

B --- FCross Pathway Link


23Understanding Gene Function through Modifier Screens:Understanding Gene Function through Modifier Screens:Synthetic Genetic Array (SGA) analysis in Synthetic Genetic Array (SGA) analysis in S. cerevisiaeS. cerevisiae

arrayed library of ~4800 viable gene deletions

gene ‘x’deleted

Tong et al. (2001)

systematic generationof double mutants


24Understanding Gene Function through Modifier Screens:Understanding Gene Function through Modifier Screens:Synthetic Genetic Array (SGA) analysis in Synthetic Genetic Array (SGA) analysis in S. cerevisiaeS. cerevisiae

Tong et al. (2001)

gene ‘x’deleted

systematic generationof double mutants


25Synthetic Genetic Array (SGA) analysis in Metazoans?Synthetic Genetic Array (SGA) analysis in Metazoans?


26Synthetic Genetic Array (SGA) analysis in Metazoans?Synthetic Genetic Array (SGA) analysis in Metazoans?


27

high degree of biological conservation to other animals

small (~1 mm)

hermaphroditic

three day life cycle

the path to many fundamental discoveries

(-) control GFP(dsRNA)

The Nematode Worm The Nematode Worm Caenorhabditis elegansCaenorhabditis elegans

RNA interference (RNAi)

Roy Lab, Univ. Toronto

The Interaction Matrix of ~56,000 Growth ScoresThe Interaction Matrix of ~56,000 Growth Scores

The Interaction Matrix of ~56,000 Growth ScoresThe Interaction Matrix of ~56,000 Growth Scores


30The SGI Network: 1246 Interactions among 461 GenesThe SGI Network: 1246 Interactions among 461 Genes

1246 synthetic genetic interactions- 842 (68%) between query genes and one of the genes in the signaling set - 421 (34%) between query genes and one of the genes in the LGIII set


31

++

Co-expression Physical Interactions

Phenotype

Previously Identified NetworksPreviously Identified Networks

Genetic Interactions

Creating a Superimposed NetworkCreating a Superimposed Network

SGI NetworkSGI Network

==

Superimposed NetworkSuperimposed Network


32

Superimposed NetworkSuperimposed Network

SGICo-expressionWorm PhenotypeProtein-proteinWorm GeneticSGI Gene“The bar-1 Subnetwork”

Mining the Superimposed for Multiply-supported SubnetworksMining the Superimposed for Multiply-supported Subnetworks

verified interaction


33

N2; Ø(RNAi) (DIC) N2; T20B12.7(RNAi) (DIC)

N2; Ø(RNAi) (Nile Red) N2; T20B12.7(Nile Red)

Genes in the Genes in the bar-1 bar-1 Subnetwork have a Shared PhenotypeSubnetwork have a Shared Phenotype


34

N2; Ø(R

NAi) F2

bar-1

; Ø(R

NAi) F2

N2; Y48

E1B.5

(RNAi)

F1

N2; mrp

-5(R

NAi) F1

N2; F29

C12.4

(RNAi)

F1

N2; ZC39

5.10

(RNAi)

F2

N2; lin-

2(RNAi)

F2

N2; B04

32.3(

RNAi) F1

N2; T20

B12.7(

RNAi) F2

N2; efl-

1(RNAi)

F2

N2; lin-

39(R

NAi) F2

N2; C27F

2.10

(RNAi)

F2

N2; lin-

35(R

NAi) F2

N2; ogt

-1(R

NAi) F2

N2; prx

-5(R

NAi) F1

N2; T09

A5.5(

RNAi) F1

N2; ubc

-18(R

NAi) F1

N2; lin-

23(R

NAi) F1

N2; F54

C9.6(

RNAi) F1

N2; exo

-3(R

NAi) F1

N2; lin-

7(RNAi)

F2

N2; T01

E8.6(

RNAi) F1

0

0.2

0.4

0.6

0.8

1

1.2

Genotype

Nor

mal

ized

N2

Val

ues

Genes in the Genes in the bar-1 bar-1 Subnetwork have a Shared PhenotypeSubnetwork have a Shared Phenotype

75% of genes in subnetwork have altered fat levels.


35

Analysis of the SGI NetworkAnalysis of the SGI Network

• Can we assign function to uncharacterized genes based on their neighborhood within the network?

• How do genetic interactions contribute to our understanding of the system?

• Can we assign function to uncharacterized genes based on their neighborhood within the network?

• How do genetic interactions contribute to our understanding of the system?


36

• Identified 343 subnetworks• 47% are enriched for a specific GO category

• 46 subnetworks are bridged by SGI links• 19-fold enriched

or

SGICo-expressionWorm PhenotypeProtein-protein

What is the Connectivity of Synthetic Genetic Links within What is the Connectivity of Synthetic Genetic Links within the Superimposed Network?the Superimposed Network?


37SGI Interactions Significantly Bridge SubnetworksSGI Interactions Significantly Bridge Subnetworks


38

Analysis of gene pairs tested for interaction in both worm and yeast

Analysis of subnetwork bridging by worm and/or transposed yeast interactions

wormyeast

OR

OR

Is the Connectivity of Genetic Networks Conserved?Is the Connectivity of Genetic Networks Conserved?


39

No evidence for bridging conservation

WORM


40

YEAST

No evidence for bridging conservation


41

Summary• Causal order from phenotypes under

knock-down

• Genome-wide interactions reveal gene functions.

• Pathway coordination may be evolvable.


42

Current Directions

• Predict drug targets from knock-down signatures

• Develop a tool for visualization and search of integrated data


43

Directions: Predict Drug Targets

• Redundancy of pathways gives synthetic lethal signature

• Compare knock-down profiles of gene A with drug X

Lokey Lab (UCSC), Davis Lab (Stanford)


44

probability genesmatch drug signatures


knock-downsensitivies

to drug

are specific pathwayspredicted?


45


See poster by Alex Williams


46

Directions: Interaction Browser

• Physical interactions

• New, high-throughput datasets

• Browser with “tracks” of interactions

• Public, 0-setup


47

AcknowledgementsAcknowledgementsMatt Weirauch

• Roy Lab– Alexandra Byrne

• Lee Lab

• Yildiz Lab

• Davis Lab– Bob St. Onge

• Stuart Lab– Martina Koeva

• Funding


48

Questions?

Interaction Browser Demo


49

Supplementary Material


50

Principle #2

Genes self assemble into modular subcomponents

0

10

20

30

40

50

605 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

105

110

115

Core Size

Per

cen

t o

f C

ore

s

Network

Random


51

Principle #3

Coordinated activity is a signature of gene function

proliferation

transcription

ribosomebiogenesis

ribosomalsubunits

respirationprotein modification

secretion

fatty acidmetab.tissue growth

neuronal

immune response

development /hox genes

cell polarity,cell structure

Newly evolved


52


53

More Projects

• Predict cancer signaling pathway from knock-down data (w/ Norm Lee at TIGR)

• Gene isoform networks to capture alternative splicing (w/ Manny Ares)

• Predict drug targets from synthetic lethal networks (w/ Scott Lokey)


54

• Elucidation is hierarchical, but no reason network should be!


55

Colon Cancer Networks


56


57

Case I


58

Case II


59

Case III


60

Probability Model


61

Probability Model


62

Colon cancer network


63

Network link probabilies correlate with confidence


64

Network

Linksa Nodesb Supported Linksc

Genetically-Supported Links (A)d

Genetically-Supported Links (B)e

Physically-Supported Linksf

Co-Exp.-Supported Linksg

Co-Phen.-Supported Linksh

Superimposed network 75,283 7,825 929 (7.2) na na na na na

wSGI 1,246 461 63 (2.0) 43 (1.6) 53 (1.8) 9 (5.6) 2 (9.0) 4 (5.9)*

Lehner 341 161 25 (5.5) 13 (10.8) 23 (7.3) 3 (22.7) 1 (17.9) 1 (30.3)

Fine genetic interactions 2,279 1,022 152 (4.6) na 48 (1.7) 61 (27.8) 23 (36.1) 22 (20.2)

Transposed SGA 7,527 426 66 (2.3) 5 (4.5) 5 (3.2)* 43 (2.2) 14 (3.0) 4 (1.3)*

Interolog 12,796 4,339 723 (9.9) 61 (27.8) 110 (4.8) na 577 (14.6) 42 (3.9)

C. elegans protein interaction 3,967 2,624 27 (3.7) 7 (10.6) 10 (4.2) na 13 (3.8) 5 (3.4)*

Eukaryotic co-expression 43,363 5,232 695 (11.8) 23 (36.1) 40 (7.2) 577 (14.6) na 84 (6.1)

C. elegans co-phenotype 8,862 913 153 (5.2) 22 (20.2) 30 (6.1) 42 (3.9) 84 (6.1) na


65

Zhong, W. & Sternberg, P. W. Genome-wide prediction of C. elegans genetic interactions. Science 311, 1481-1484 (2006).

• Combined interactome, gene expression, phenotype, functional annotation data

• Yeast, fly, and worm

• Used a training set of 1816 previously reported genetic interactions and 2878 P2P interactions.

• Assigned each type of evidence a weighted predictive score

• Gave a prediction score to each possible pair of genes

• Predicted 18,183 interactions among 2254 genes

• Validated 12 of 49 novel predicted interaction with let-60• Validated 2 of 6 novel predicted interactions with itr-1

Oct 12, 2007 Research Review Day 1 Josh Stuart, Ph.D. Biomolecular Engineering UCSC Research Review Day 2007 Biological Discovery From Genetic Network.

Documents

ucsc slide

gene function slide

gwumc slide

microarray slide

knock downs

genome hair neuron slide

cellmigration slide

software of life slide