Transcript Genetic Analysis of Genome-wide Variation in Human Gene

Genetic Analysis of Genome-wide
Variation in Human Gene Expression
Morley M. et al. Nature 2004,430: 743-747.
Yen-Yi Ho
Outline
•
•
•
•
•
•
Introduction
Data
Method: Linkage analysis
Results
Discussion
Comments
Introduction
Goal : Identify loci associated with variation in
expression levels
Nucleus
regulators
Genomic DNA
mRNA
Target
mRNA
Cis and Trans regulation
Trans-regulator
Cis-regulator
Target gene expression
phenotype
Data
Centre d'Etude du Polymorphisme Humain (CEPH) families are
Utah residents with ancestry from northern and western
Europe.
• 14 families with genotype and expression data available
for all parents and a mean of eight offspring (range 7-9)
• 2,756 autosomal SNP genetic loci (100kb within
cluster, 3 Mb between cluster).
• Gene expression phenotypes
~8,500 gene expression phenotypes in immortalized
B cells using Affymetrix Genome Focus Array.
Expression intensity was scaled to 500 and
transformed by log2.
• 3,554 most variable expression phenotypes are
selected (between > within variation).
• Using CEPH unrelated individuals (94
grandparents), two array replicates per
individual was performed. The within individual
variation was indicated by the mean of variance
of array replicates.
Method: Linkage analysis
A1 A2  A3A4
A1 A3
A1 A3
IBD=2
IBD: identical-by-descent
A1 A2  A3 A4
A1 A3
IBD=1
A1 A4
A1 A2  A3 A4
A1 A3
A2 A4
IBD=0
(Yi1  Yi 2 )2
(Yi1  Yi 2 )2
E (DIS)   0  1 I(IBD=2) +  2 I(IBD=1)
H 0 : 1   2  0
F
RSS0  RSS1
~ F2,n 1
RSS1
DIS: (Yi1  Yi 2 ) 2 , the distance of gene expression between sib 1 and 2.
n: the total number of sibpairs
For a particular target gene expression
15
t-statistics
10
5
SNP1
2 3
4 5
Genetic Locus
Results
Criteria 1 : t > 5 (P-value < 4.3 107, LOD > 5.3) : 142 expression phenotypes
have at least one significant regulator.
Criteria 2 : t > 4 (P-value < 3.7 10 5 , LOD > 3.4): 984 expression phenotypes
have at least one significant regulator.
Cis and trans- regulation
Under criteria 1,
• 27/142 (19%) expression phenotype have only a
single cis-regulator.
• 110/142 (77.5%) expression phenotype have only
a single trans-regulator.
• 2 /142 have a cis and a trans-acting regulator
• 3 /142 gene expression have two trans-acting
regulator
Under criteria 2,
164 / 984 (16%) has multiple regulators
T-statistic
1
2
Gene
expression 3
.
“Target”
.
.
4 7
…1
Criteria 2
1 3
3554
t > 3.4
0 0
...
...
8 6
1 1
SNP 1
2
3
…0
1 1
…..2756
SNP 1
2
3
…..2756
Genetic Locus “regulator”
5 Mb window (total 491 windows)
Master regulator
31
25
14q32
20q13
Divide the autosomal genome into 491 windows of 5 Mb, and count the number of
regulators in the regions under criteria 2 (total 984 phenotypes with significant
linkages).
Co-regulation
• Use the gene expression levels of 94 CEPH
grandparents
• Hierarchical clustering was performed and group
genes by the correlation of the 31 target gene
expression levels
• Permutation test was used to determine the
significant level of pair-wise correlation.
14 / 31
Population-based association analysis for cis-regulators
(SNP regulator)
Discussion
• The study applied genome-wide mapping
method to identify the chromosomal regions
regulate to the gene expression phenotypes.
• This type of study has the potential to
uncover complicated transcriptional control.
• Cis-, trans-acting and master regulators
were discovered.
• The linkage results are reliable as verified
by association study and qRT-PCR.
Comments
• In this study, gene expression measured in
immortalized B cells may be very different from
the expression of human B cells in the blood.
• Co-regulated genes and the pathways that
connect genes are identified.
• We would be even more interested in utilizing the
data to improve our understanding of human
disease.
Genotype
Gene expression
Phenotype
• Candidate regions have cis-effects.
• Different phenotype / expression
signatures associated with different
“regulators”.
Statistical design and analysis issues
Design:
• Choice of relative type or pedigree in humans.
• Choice of tissue and timing of mRNA sampling.
Analysis:
• Multiple testing: linkage location, transcripts.
• Regulatory hotspots: methods to find master
regulatory loci.
• Regulatory networks: searching for small sets of
Co-regulated transcripts.
Reference
1.
2.
3.
4.
5.
Genetic analysis of genome-wide variation in human gene
expression. Moley M., Molony C.M, Teresa M. Weber T.M. et al.
Nature 430:743-747 (2004).
Genetics of gene expression surveyed in maize, mouse, and man.
Schadt E.E., Monks S.A., Drake T.A. et al. Nature 422: 297-302
(2003).
Mapping expression in randomized rodent genomes. Broman K.W.
Nature Genetics 37: 209-210 (2005).
Natural variation in human gene expression assessed in
lymphoblastoid cells. Cheung V.G., Conlin L.K., Weber T.M. et al.
Nature Genetics 33: 422-425 (2003).
Mapping determinants of human gene expression by regional and
genome-wide association. Cheung V.G., Spielman R.S., Ewens
K.G. et al. Nature 437: 1365-1369 (2005).
Question?