Transcript Microarray Analysis & Functional Genomics

Functional Genomics in
Evolutionary Research
What Is Microarray Technology?
High throughput method for measuring
simultaneously, mRNA abundances for
thousands of genes.
Thousands of probes or features
adhered to a solid substrate at
known x,y coordinates.
Probes:
Spotted cDNA ~ 200 bp
Oligo = 25 to 60 bp
Why Is Microarray Technology Important?
From NSF Program Announcement: Environmental Genomics
How Do Microarrays Work?
Hybridization Technique
- RNA targets isolated from a cell line or tissue of interest
are labeled and hybridized to the probes.
- Label intensity at a given location on the substrate
correlates with the amount of target for a given mRNA
(gene) present in the sample.
Differentially expressed Genes:
Identified statistically (e.g. t-test)
by comparing control vs experimental.
The Burden of Multiple Testing
A given microarray may have over 40,000 probes!!!
This means that you may run as many as 40,000
statistical tests.
If you reject a null hypothesis when P < 0.05, then 5%
of the time you are rejecting true null hypotheses.
If you run 40,000 tests, then by chance alone you will
reject ~ 40,000 x 0.05 = 2000 true null hypotheses
(i.e., you will have ~ 2000 false positives)
Sources of Variation in Microarray Experiments
Biological
Technical (Bad)
(1) Experimental Treatments
(1) RNA quality
(2) Individual variance... may or
may not be good
(2) Dye biases
(3) Nonspecific hybridization
Paralogs of gene families
(3) Stochasticity during scanning,
image processing
(5) Errors during probe synthesis or
deposition
(6) Stochasticity in labeling targets
What gene expression changes are
associated with the evolution of
paedomorphosis?
Example Design
(1) Sample tissue from 15 time
points (x), including an early
reference (R) time point.
(2) Compare expression for
each time point and the reference
on a DNA chip.
Metamorphic Life Cycle
Larval
Metamorphosis
Adult
R x x x x
xxxxxx
x x x x
R x x x x
xxxxxx
x x x x
Larval
Metamorphosis?
Adult
Paedomorphic Life Cycle
(3) Quantify relative expression
of each gene across all DNA chips.
(2 life cycles x 3 tissues x 14
timepoints)
(7)Verify results by rt-PCR and
analyze candidates in thyroid
hormone-induced paedomorphs.
(6) Compare gene expression
profiles among life cycles and
tissues to identify differentially
expressed genes.
(4) Model gene expression to determine
how genes are expressed temporally
within life cycle cycles for each tissue.
Visualization & Categorization
Can be done for genes and/or arrays... Options Include a variety of
multivariate and pattern matching techniques including the methodologies
listed below
Quadratic Regression
Liu et al. 2005... From the Stromberg Group
here at UK
Principal Component
Analysis
Clustering
Heat maps
Gene Ontology & Biological Relevance
• Microarray datasets can be overwhelming because they contain
A LOT of information
• Even experts on a system can be overwhelmed by the number of
genes that are differentially regulated in some experiments
• Having a standardized nomenclature that places a gene into one
or more biological contexts can be invaluable for functional
grouping (previous grouping techniques were irrespective of
biological information)
Gene Ontology is a standardized
hierarchical nomenclature that
classifies genes under three
broad categories