A Quantitative Overview to Gene Expression Profiling in Animal Genetics

A Quantitative Overview to Gene Expression Profiling in Animal Genetics

Armidale Animal Breeding Summer Course, UNE, Feb. 2006

Co-Expression

Reverter et al. 2005Bioinformatics 21:1112

Validation of alternative methods of datanormalization in gene co-expression studies



Co-ExpressionMotivation

Statistics are like a bikini,what they reveal is suggestive,but what they conceal is vital.

Aaron Levenstein



Co-ExpressionMotivation

Correlations Are Dangerous!

##

##

# ###

##

#

#

#

##

#

##

##

# ###

##

#

#

#

#

#Exp 1

Exp 2

Gene 1

Gen

e 2

Gene 1

Gen

e 2

##

##

#

#

#

#

#

##

#

#

#

#

#

#

#

#

#

#

#

#

#

## #

#

##

#

#

#

#

#

##

#

#

#

#

#

#

#

#

#

#

#

#

#

## #

#

##

#

#

#

#

#

##

#

#

#

#

#

#

#

#

#

#

#

#

#

##

2

1)(

2

n

rrSE



Co-ExpressionIntroduction

Validation of alternative methods of data normalization in gene co-expression studies. Bioinformatics 2005, 21:1112

A. Reverter, W. Barris, S.M. McWilliam, K.A. Byrne, Y.H. Wang, S.H. Tan, N. Hudson, and B.P. Dalrymple

Experimenta Hybs. Cond. Signalsb

N Mean STD

1. Two breeds by two diets 7 4 193,175 7.61 2.99

2. Three diets 14 3 361,320 8.52 2.84

3. Two diets at three ages 24 6 801,807 11.60 1.89

4. Two breeds at three ages 18 6 459,978 8.34 3.28

5. Two fat treatments at two ages 15 4 418,817 7.04 3.31

Expression of each clone (gene) across 23 conditions

A Quantitative Overview to Gene Expression Profiling in Animal GeneticsCo-Expression

Introduction


1

2

3

4

5



Co-ExpressionMethods

Nine Methods of Data Normalization

######################

Clone i

Exp

erim

ent 1

( E

1, E

1)E

xper

imen

t 2

(E

2, E

2)

Cond

1

2

3

4

5

(Ci, Ci)

1x

2x

3x

4x

5x

1E1E1 /)x(

1E1E2 /)x(

1E1E3 /)x(

2E2E4 /)x(

2E2E5 /)x(

ii CC1 /)x(

ii CC2 /)x(

ii CC3 /)x(

CiCi4 /)x(

CiCi5 /)x(

1. RMNA 2. RMCE 3. RMEC

















93.142.960.8

55.180.900.10






17.232.960.8

95.145.900.10






Expected to be the equal!





eTv

Ta

Ta

Tg VVDDAAGGXMVNY ,~

Log2Intensities

Comparison GroupArray|Block|Dye

(FIXED)Main GeneEffect(RANDOM)

Gene xVariety(RANDOM)

Residual(RANDOM)

DE GenesGene x

Array|Block(RANDOM)

Gene x Dye(RANDOM)

Mixed-Model Equations





4. MM1NA 5. MM1CE 6. MM1EC








ct̂E

Ecc

/

ˆ t

c

cEc

/

ˆ t

cEt̂

Vector of BLUPClone x Condition

Interaction

E

EccE

/

ˆ t

c

cEcE

/

ˆ t



Co-Expression

Observed (RED) vs Expected (BLUE)Results



Co-ExpressionResults

Average across all pair-wise gene correlations (bold, italics and on the diagonal); Correlation of correlations (above diagonal); and

maximum absolute discrepancy between methods (below diagonal)

1 2 3 4 5 6 7 8 9

1. RMNA 0.53 0.20 0.98 0.11 0.12 0.11 0.06 0.07 0.06

2. RMCE 1.84 0.15 0.21 0.14 0.14 0.14 0.09 0.10 0.09

3. RMEC 0.76 1.79 0.51 0.12 0.12 0.12 0.07 0.07 0.07

4. MM1NA 1.78 1.72 1.74 0.02 0.94 0.99 0.57 0.60 0.56

5. MM1CE 1.75 1.75 1.74 0.51 0.01 0.94 0.61 0.69 0.61

6. MM1EC 1.79 1.71 1.74 0.50 0.55 0.02 0.57 0.61 0.58

7. MM5NA 1.80 1.75 1.79 1.25 1.14 1.21 0.01 0.91 0.99

8. MM5CE 1.78 1.74 1.77 1.19 1.09 1.19 0.59 0.01 0.90

9. MM5EC 1.80 1.75 1.79 1.25 1.13 1.20 0.55 0.66 0.01




Redundant genes (ie. those with many probes on the array) allow for Validation



Co-ExpressionResults Redundant genes allow for Validation




624

Ge

ne

Co

-Exp

ress

ion

(Alp

ha

betic

al o

rder

)

Robisomal Proteins



Co-Expression

Brian Sigrid

Siok Kwee Keren

Applications



Co-ExpressionApplications

False Discovery Rate as a Function of Correlation




DLK1

COL3A1

LUM

GNAI2

DCAL1

CIDEC

KARS

UBA52

TNFRSF21

0.584

0.602

0.611

0. 611

0.545

-0.639

-0.658

-0.697

Callipyge Mutation




DLK1

COL3A1

LUM

GNAI2

DCAL1

CIDEC

KARS

UBA52

TNFRSF21

DES

FN1

CTSF

SPARC

CAV1

ACTA1

TM4SF2

0.853

0.880

0.584

0.602

0.611 0. 736

0. 611

0.545

-0.639

-0.658

-0.697

-0.822

0.754

0.747

0.767

0.743

-0.741

-0.738

0. 799

Callipyge Mutation




DLK1

COL3A1

LUM

GNAI2

DCAL1

CIDEC

KARS

UBA52

TNFRSF21

DES

FN1

CTSF

SPARC

CAV1

ACTA1

TM4SF2

MYBPC2

MYL1

LDHA

PGK1

CASQ1

0.883

0.889

0.889

0.892

0.893

0.853

0.880

0.584

0.602

0.611 0. 736

0. 611

0.545

-0.639

-0.658

-0.697

-0.822

0.754

0.747

0.767

0.743

-0.741

-0.738

0. 799

Callipyge Mutation




DLK1

COL3A1

LUM

GNAI2

DCAL1

CIDEC

KARS

UBA52

TNFRSF21

DES

FN1

CTSF

SPARC

CAV1

ACTA1

TM4SF2

MYBPC2

MYL1

LDHA

PGK1

CASQ1

0.883

0.889

0.889

0.892

0.893

0.853

0.880

0.584

0.602

0.611 0. 736

0. 611

0.545

-0.639

-0.658

-0.697

-0.822

0.754

0.747

0.767

0.743

-0.741

-0.738

0. 799

0.911

ITM2B

ENO3

PYGM-0.899 0.916

0.972

GPI0.897

CKM0.910

Callipyge Mutation

A Quantitative Overview to Gene Expression Profiling in Animal Genetics

Documents

gene expression profiling

clone gene

gene coexpression studies

clone iexperiment

fat treatments