1. Gene Expression BMI 731 week 5
Catalin Barbacioru
Department of Biomedical Informatics
Ohio State University

2. Thesis: the analysis of gene expression data is going to be big in 21st century statistics
Many different technologies, including
High-density nylon membrane arrays
Serial analysis of gene expression (SAGE)
Short oligonucleotide arrays (Affymetrix)
Long oligo arrays (Agilent)
Fibre optic arrays (Illumina)
cDNA arrays (Brown/Botstein)*

3. Total microarray articles indexed in Medline
100
200
300
400
500
600
(projected)
Year
Number of papers

4. Common themes
Parallel approach to collection of very large amounts of data (by biological standards)
Sophisticated instrumentation, requires some understanding
Systematic features of the data are at least as important as the random ones
Often more like industrial process than single investigator lab research
Integration of many data types: clinical, genetic, molecular…..databases

5. Biological background
G U A A U C C
RNA polymerase
mRNA
Transcription
DNA
G T A A T C C T C
| | | | | | | | |
C A T T A G G A G

6. Idea: measure the amount of mRNA to see which genes are being expressed in (used by) the cell.
Measuring protein might be better, but is currently harder.

7. Reverse transcription
Clone cDNA strands, complementary to the mRNA
mRNA
G U A A U C C U C
Reverse transcriptase
cDNA
T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G
C A T T A G G A G

8. cDNA microarray experiments
mRNA levels compared in many different contexts
Different tissues, same organism (brain v. liver)
Same tissue, same organism (ttt v. ctl, tumor v. non-tumor)
Same tissue, different organisms (wt v. ko, tg, or mutant)
Time course experiments (effect of ttt, development)
Other special designs (e.g. to detect spatial patterns). 4

9. DNA microarrays represent an important new method for determining the complete expression profile of a cell.
Monitoring gene expression lies at the heart of a wide variety of medical and biological research projects, including classifying diseases, understanding basic biological processes, and identifying new drug targets.

10. Affymetrix® Instrument System
Platform for GeneChip® Probe Arrays
Integrated
Exportable
Easy to use
Versatile

11. Photolithography

12. Synthesis of Ordered Oligonucleotide Arrays
O O O O O
Light
(deprotection)
HO HO O O O
T T O O O
T T C C O
C A T A T
A G C T G
T T C C G
Mask
Substrate
T –
C –
REPEAT
Light removes protecting groups at defined positions. Single nucleotide washed over the chip, binds where the protecting group removed. Through successive steps, any sequence can be built up in any position on the chip. The number of steps corresponds with length of oligo, so can increase # of genes without # of steps

13. Affymetrix GeneChip arrays

14. GeneChip® Probe Arrays*
Hybridized Probe Cell
GeneChip Probe Array
Single stranded,
labeled RNA target
Oligonucleotide probe
24µm
Millions of copies of a specific
oligonucleotide probe
1.28cm
GENECHIP PROBE ARRAYS
The core of the platform is our unique arrays
Oligonucleotides synthesized de novo (photolithography & combinatorial chemistry)
Currently 65,000 (50 micron features) to 250,000 (24 micron) different oligos on commercially available products
Each oligo represented in 107 to 108 full-length copies
>200,000 different
complementary probes
Image of Hybridized Probe Array

16. Analysis of expression level from probe sets
A single, contiguous gene set for the rat B-actin gene.
Each pixel is quantitated and integrated for each oligo feature (range 0-25,000)
Perfect Match (PM)
Mis Match (MM) Control
log(PM / MM) = difference score
All significant difference scores are averaged to create “average difference” = expression level of the gene.

17. Expression screening by GeneChip
• each oligo sequence (20-25 mer) is synthesized
as a 20 µ square (feature)
• each feature contains > 1 million copies of the oligo
• scanner resolution is about 2 µ (pixel)
• each gene is quantitated by oligos and
compared to equal # of mismatched controls
• 22,000 genes are evaluated with 20 matching oligos
and 10 mismatched oligos = 480,000 features/chip
• 480,000 features are photolithographically synthesized
onto a 2 x 2 cm glass substrate

18. Affymetrix GeneChip arrays
Global views of gene expression are often essential for obtaining comprehensive pictures of cell function.
For example, it is estimated that between 0.2 to 10% of the 10,000 to 20,000 mRNA species in a typical mammalian cell are differentially expressed between cancer and normal tissues.
Whole-genome analyses also benefit studies where the end goal is to focus on small numbers of genes, by providing an efficient tool to sort through the activities of thousands of genes, and to recognize the key players.
In addition, monitoring multiple genes in parallel allows the identification of robust classifiers, called "signatures", of disease.
Global analyses frequently provide insights into multiple facets of a project. A study designed to identify new disease classes, for example, may also reveal clues about the basic biology of disorders, and may suggest novel drug targets.

19. cDNA microarrays
In ‘‘spotted’’ microarrays, slides carrying spots of target DNA are hybridized to fluorescently labeled cDNA from experimental and control cells and the arrays are imaged at two or more wavelengths
Expression profiling involves the hybridization of fluorescently labeled cDNA, prepared from cellular mRNA, to microarrays carrying thousands of unique sequences.
Typically, a set of target DNA samples representing different genes is prepared by PCR and transferred to a coated slide to form a 2-D array of spots with a center-to-center distance (pitch) of about 200 μm, providing a pan-genomic profile in an area of 3 cm2 or less.
cDNA samples from experimental and control cells are labeled with different color fluors (cytochrome Cy5 and Cy3) and hybridized simultaneously to microarrays, and the relative levels of mRNA for each gene are then determined by comparing red and green signal intensities

20. cDNA microarrays Scanning Technology
Microarray slides are imaged with a modified fluorescence microscope designed for scanning large areas at high resolution (arrayWoRx, Applied Precision, Issaquah, WA, Affymetrix).
Fluorescence illumination are obtained from a metal halide arc lamp focused onto a fiber optic bundle, the output of which is directed at the microarray slide and emission recorded through a microscope objective (Nikon) onto a cooled CCD (charge-coupled device) camera.
Interference filters are used to select the excitation and emission wavelengths corresponding to the Cy3 and Cy5 fluorescent probes (Amersham Pharmacia).
Each image covered a 2.4 x 2.4 mm area of the slide at 5-μm resolution. To scan the entire microarray, a series of images (‘‘panels’’) were acquired by moving the slide under the microscope objective in 2.4-mm increments.

22. 16-bit TIFF files (Rfg, Rbg), (Gfg, Gbg) R, G Biological question
Differentially expressed genes
Sample class prediction etc.
Experimental design
Microarray experiment
16-bit TIFF files
Image analysis
(Rfg, Rbg), (Gfg, Gbg)
Normalization
R, G
Estimation
Testing
Clustering
Discrimination
Biological verification
and interpretation

23. Some statistical questions
Image analysis: addressing, segmenting, quantifying
Normalisation: within and between slides
Quality: of images, of spots, of (log) ratios
Which genes are (relatively) up/down regulated?
Assigning p-values to tests/confidence to results. 4

24. Some statistical questions, ctd
Planning of experiments: design, sample size
Discrimination and allocation of samples
Clustering, classification: of samples, of genes
Selection of genes relevant to any given analysis
Analysis of time course, factorial and other special experiments…..…...& much more. 4

25. Some bioinformatic questions
Connecting spots to databases, e.g. to sequence, structure, and pathway databases
Discovering short sequences regulating sets of genes: direct and inverse methods
Relating expression profiles to structure and function, e.g. protein localisation
Identifying novel biochemical or signalling pathways, ………..and much more. 4

26. Part of the image of one channel false-coloured on a white (v
Part of the image of one channel false-coloured on a white (v. high) red (high) through yellow and green (medium) to blue (low) and black scale

27. Does one size fit all?

28. Segmentation: limitation of the fixed circle method
SRG
Fixed Circle
Inside the boundary is spot (foreground), outside is not.

29. Some local backgrounds
Single channel
grey scale
We use something different again: a smaller, less variable value.

30. Quantification of expression
For each spot on the slide we calculate
Red intensity (PM) = Rfg - Rbg
fg = foreground, bg = background, and
Green intensity (MM) = Gfg - Gbg
and combine them in the log (base 2) ratio
Log2( Red intensity / Green intensity)
Log2( PM / MM)

31. Gene Expression Data = slide 1 slide 2 slide 3 slide 4 slide 5 …
On p genes for n slides: p is O(10,000), n is O(10-100), but growing,
Slides
slide 1 slide 2 slide 3 slide 4 slide 5 …
Genes 3
Gene expression level of gene 5 in slide 4 =
Log2( Red intensity / Green intensity)
These values are conventionally displayed
on a red (>0) yellow (0) green (<0) scale.

33. The red/green ratios can be spatially biased.
Top 2.5%of ratios red, bottom 2.5% of ratios green

34. Affymetrix vs. cDNA Arrays
Affy Strengths:
- highly reliable: synthesized in situ
- highly reproducible from run to run
- no clone maintenance or ‘drift’
- sealed fluidics and controlled temperature
- standardized chips increase database power
- excellent scanner
- complex, but very reliable labelling
- excellent cost/benefit ratio
- amenable to mutation and SNP detection

35. Affymetrix weaknesses/limitations
not easily customized: $300K/chip
high labeling cost $170/chip
high per chip cost $350 to $1850
limited choice of species
requires knowledge of sequence
not designed for competitive protocols

36. Limitations to all microarrays.
dynamic range of gene expression:
very difficult to simultaneously detect low and high
abundance genes accurately
- each gene has multiple splice variants
2 splice variants may have opposite effects (i.e. trk)
arrays can be designed for splicing, but complexity ^ 5X
- translational efficiency is a regulated process:
mRNA level does not correlate with protein level
- proteins are modified post-translationally
glycosylation, phosphorylation, etc.
- pathogens might have little ‘genomic’ effect

37. Analysis
In general the expression level of individual genes is measured by log(PM/MM) or log(R/G).
Intensity-dependent normalization methods are preferred over a global methods.
To correct intensity- and dye-bias we used location and scale normalization methods, which are based on robust, locally linear fits (lowess).
Global methods use linear regression models, combined with ANOVA.

38. Normalization Why? How do we know it is necessary?
To correct for systematic differences between samples on the same slide, or between slides, which do not represent true biological variation between samples.
How do we know it is necessary?
By examining self-self hybridizations, where no true differential expression is occurring.
We find dye biases which vary with overall spot intensity, location on the array, plate origin, pins, scanning parameters,….

39. Analysis
Post-normalization
Pre-normalization

40. The simplest cDNA microarray data analysis problem is identifying differentially expressed genes using replicated slides
There are a number of different aspects:
First, between-slide normalization; then
What should we look at: averages, SDs, t-statistics, other summaries?
How should we look at them?
Can we make valid probability statements? 4

41. Apo AI experiment (Matt Callow, LBNL)
Goal. To identify genes with altered expression in the livers of Apo AI knock-out mice (T) compared to inbred C57Bl/6 control mice (C).
8 treatment mice and 8 control mice
16 hybridizations: liver mRNA from each of the 16 mice (Ti , Ci ) is labelled with Cy5, while pooled liver mRNA from the control mice (C*) is labelled with Cy3.
Probes: ~ 6,000 cDNAs (genes), including 200 related to lipid metabolism.

43. Which genes have changed? When permutation testing possible
1. For each gene and each hybridisation (8 ko + 8 ctl), use M=log2(R/G).
2. For each gene form the t statistic:
average of 8 ko Ms - average of 8 ctl Ms
sqrt(1/8 (SD of 8 ko Ms)2 + (SD of 8 ctl Ms)2)
3. Form a histogram of 6,000 t values.
4. Do a normal q-q plot; look for values “off the line”.
5. Permutation testing (next lecture).
6. Adjust for multiple testing (next lecture). 9

44. Histogram & normal q-q plot of t-statistics
ApoA1

45. Patterns, More Globally...
Can we identify genes with interesting patterns of expression across arrays?
Two approaches:
1. Find the genes whose expression fits specific, predefined patterns.
2. Perform cluster analysis - see what expression patterns emerge.

48. The 16 groups systematically arranged (6 point representation)