1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi...

27
1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci

Transcript of 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi...

Page 1: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

1

CIS 4930/6930 – Recent Advances in Bioinformatics

Spring 2014

Network construction from RNAi data

Tamer Kahveci

Page 2: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Signaling Networks

2

MAPK network

Page 3: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Signal reachability

3

Receptor Reporter

Luciferase

Page 4: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Signaling and RNA Interference

4

Receptor Reporter

Luciferase

X Not critical

X Critical

Page 5: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Signaling Network Reconstruction from RNAi data

5

Receptor Reporter

Not criticalCritical

Page 6: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

RNAi data and Reference Network

6

Receptor Reporter

Not criticalCritical

Reference network

Inse

rt

Delete

Not consistent !Consistent

!

Page 7: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Overview

7

GR = (VR, ER)

Reference network

Constraints

1 10 0 0

GT = (VT, ET)

Target network

1 0

SiNeC (Signal Network Constructor)

S-SiNeC (Scalable Signal Network Constructor)

Giv

en Find

Goal: Minimize the number of edit

operations to make the reference

consistent.

NP-Complete !

Page 8: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

SiNeC algorithm

Three steps

1. Order the critical genes left to right based on the topology of GR. [Sloan, 1986]

– v1, v2, …, vc

2. Edge deletion phase

3. Edge insertion phase

8

Page 9: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Step 1: Order critical genes

9

Receptor Reporter3

1

2

Prioritize based on distance to the reporter + degree

Page 10: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Step 2: Edge deletion

10

Purpose: Eliminate detours around critical genes

Receptor Reportervi vkvj

• Find all (undesirable) paths between non-consecutive critical genes.

• i.e., Paths which go through only noncritical genes• Edges are weighted with the number of such paths they

belong to.• Remove greedily starting from the largest weight until al

paths are disrupted.

Bypassed !!!

Page 11: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Step 3: Edge insertion

11

Purpose: Make sure that critical are connected + noncritical genes are consistent

Receptor Reportervi-1 vi+1vi

Insert an edge from vi-1 to vi if 1. There is no path from vi-1 to vi.2. There is a noncritical gene on all paths from vi-1

to vi.

Page 12: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Overview

12

GR = (VR, ER)

Reference network

Constraints

1 10 0 0

GT = (VT, ET)

Target network

1 0

SiNeC (Signal Network Constructor)

S-SiNeC (Scalable Signal Network Constructor)

Giv

en Find

Finding all the paths can be to

o time

consuming for la

rge networks

Page 13: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC algorithm

13

Edge insertion0 0 0 None

0 0 1 None

0 1 0 None

0 1 1 A1

1 0 0 A2 + A3 + A4

1 0 1 A2 + A4

1 1 0 A3 + A4

1 1 1 A4

Critic

alLe

ft rea

chab

le

Rig

ht rea

chab

le

Edge deletion

Reference network

vs vtvi

Page 14: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC: Edge insertion (A1)

14

Reference network

vs vtvi

L R

Purpose: Make sure that noncritical genes are consistent

Page 15: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC: Edge insertion (A2)

15

Reference network

vs vtvi

L R

Purpose: Make sure that critical genes are left reachable

Page 16: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC: Edge insertion (A3)

16

Reference network

vs vtvi

L R

Purpose: Make sure that critical genes are right reachable

Page 17: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC: Edge insertion

17

L/R e1 e2 e3

1 X

2 X X

3 X X

4 X X

5 X X

6 X X

7 X

8 X

Page 18: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

S-SiNeC: Edge deletion (A4)

18

Reference network

vs vtvi

L R

Purpose: Make sure that no detours exist around critical genes

Solve minimum cut between L &

R

Page 19: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Dataset

19

• Reference networks are obtained by random edge shuffling at 5% to 40% mutation rates.

• 200 references per target network & per mutation rate.

Page 20: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Average distance to the true network

20

Page 21: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Accuracy based on edge class

21

vs vt

Hot

Cold

Page 22: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Running time results

22SiNeC > 1 hour per reference network.

Page 23: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Success rate on constraints

23

Page 24: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

24

Accuracy

Page 25: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

25

Functional Enrichment of the Pathway

Page 26: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Last Remarks

• Constructing very large signaling networks from RNAi data is possible in practical running time.

• Both SiNeC and S-SiNeC are robust to errors in reference network.

• We recommend

– S-SiNeC for very large OR dense networks.

– SiNeC otherwise.26

Page 27: 1 CIS 4930/6930 – Recent Advances in Bioinformatics Spring 2014 Network construction from RNAi data Tamer Kahveci.

Acknowledgements

27

CCF - 0829867 IIS - 0845439260429