Download Gene Expression Analysis: Differential Expression & Statistical Inference and more Study Guides, Projects, Research Computer Science in PDF only on Docsity!

Project Introduction:Gene Expression Analysis

Utah State UniversityBioinformatics: Problems and SolutionsSummer 2006

Traditional questions & extensions „^

Which genes are really changing expression levelbetween conditions (disease, tissue, genotype,etc.)? „^

How much (and in which direction) are they reallychanging? „^

Where do these genes lie? (chromosomal location) „^

How are these genes different from the others?(gene ontology)

A generalized t-test in a linear model (limma) „^

For gene k under “treatment” j on array i:

„^

What if there are more covariates than justtreatment? –

use matrix notation for convenience:

[^

]^

2

(^1) ,

(^0) ,

,^

k

ijk

ijk

jk k

k

ijk^

Var

T

Y

σ

ε

ε

β

β^

=

=

expressionlevel(log scale)

treatmenteffect (DE)

treatment level (could bemore than just 2 levels)

[^

]^

k

k^

X

Y E

β

covariate effects design matrix (n x m)

log-scale expression vector

Assumptions in linear model (Smyth)

[^

]

(^

k

ww

k

ww

d

k k

wk

wk

k

k

k

d k k

k k

k k wk

k wk

wk

k k

k

k

k

t

V

t

m

n

d

d

V

N

w

V

Var

Then

d.f.

resid.

covariate

For

and

and

estimates

Obtain

,

,

,

,

2 2

2 2

2

,

2 ,

,

2

σ

β

χ σ

σ σ

σ

β

σ

β

β

σ

β

σ

β

k not necessary here

Differential expression: a quick example in R

# load datalibrary(ALL);

data(ALL)

# define comparison (based

on knowledge of samples)

eset <- ALL # these are normalized expression levelssampleNames(eset)trt <- c(rep(0,95),rep(1,33)) # 0=B, 1=T# test for differential expression (DE)library(limma)design <- cbind(Intercept=1,trt=trt)fit <- lmFit(eset@exprs,design)e.fit <- eBayes(fit)# Visualize resultstop.all <- topTable(e.fit,n=nrow(eset@exprs),

coef=2,adjust="BH")

hist(top.all$P.Value,main='raw P-value')hist(top.all$adj.P.Val,main='adj. P-value')sum(top.all$adj.P.Val<0.05)

“gene” location info.

“gene” function info.

“gene” DE results

Background „

Barley (Hordeum vulgare)^ ‰

grain used for animal feed (poultry, e.g.)and human use (bread, beer) ‰ Utah is among the top 12 U.S. producers „

Several cultivars (strains or genotypes):Kindred, Peruvian, Beka, … „

Susceptible to pathogens causing leafblotch:

Septoria passerinii

&

Septoria tritici

Questions & Problems „^

Questions: ‰^

Which genes are differentially expressed betweencultivars? ‰^

Where are these genes? ‰^

How are they different from other genes?

„^

Problems: ‰^

Chromosomal location not automatically included in anannotation package for barley – but other sources possible: „^

hvuhomology package for R

„^

barleybase.org

, plantgdb.org, gramene.org

„^

NetAffx from affymetrix.com

‰^

Gene ontology information not available for all probe sets

Project description „^

Analyze these barley data, focusing on: ‰^

graphical display of results ‰^

visualization of chromosomal locations ofsignificant genes ‰^

summarization of possible over-representation ofgene ontology terms among differentiallyexpressed genes

„^

All team members will need to contribute,culminating in poster presentation