Microarrays

1
Microarrays

• By
• Heather Matthews
• Tom Shi
• Shih-Hsuan Wu

2
Agenda

• Introduction, History, Discoveries and
  Applications
• Heather
• cDNA Microarray
• Tom
• Synthetic Oligonucleotide Array
• Shih-Hsuan

3
Overview

• Biology background
• What do microarrays do?
• Why are microarrays important?
• Types of experiments done using microarrays

• Applications of microarrays
• History of microarrays
• Important experiments using microarrays
• The future of microarrays

4
Biology Background

• DNA in somatic cells, all cells except the
  gametes, is the same regardless of tissue type or
  physiological state.
• Proteins in cells can differ depending on whether
  genes are expressed in the cells and at what
  level.
• Genes are expressed at different levels in cells
  depending on the
• -tissue type
• -time in the cell cycle
• -developmental stage
• -physiological condition

5
What do microarrays do?

• Microarrays can measure the expression level of
  thousands of genes at once.
• This is called doing experiments in parallel.

6
Why would you want to know the expression level
of genes?

• Its important to know what level genes are
  expressed at in certain cells, because in a
  diseased cell, for example a cancer cell, the
  genes that are over or under expressed could
  potentially be regulated to the expression level
  of a normal cell, which could help in the
  treatment of the disease.
• Drugs could be developed to inhibit proteins that
  are too abundant in diseased cells.

7
Types of microarray experiments

• Compare/contrast expression levels
• -in disease vs. non-disease states, such as
  healthy cells vs. cancer cells.
• -in different types of tumor cells
• -in different types of tissue, such as muscle
  vs. nerve cells.

8
Type of microarray experiments continued

• Compare/contrast expression levels
• at different stages in the cell cycle
• in organisms treated with a drug vs. a control
  group
• of genes from the same type of cell in different
  environments
• of genes from the same type of cell in different
  species

9
Applications of microarrays

• Drug Discovery - Potential to identify certain
  genes that could indicate targets for drugs to
  help cure certain diseases.
• Diagnostics Diagnosis of diseases for which a
  gene mutation has been identified.
• Custom Drug Selection Alternative forms of
  genes can affect the action or metabolism of a
  drug. Specific drugs and/or dosages could be
  customized for each patient.
• Accelerate FDA approval

10
History of microarrays

• Current microarray technology is very recent.
• One of the first articles using microarrays was
  written in October of 1995 (Schena et al. 1995)
• However, quantitative methods to measure gene
  expression levels that involve hybridization have
  been around for more than 20 years.

11
Other methods used to measure gene expression
levels involving hybridization

• Northern and Southern Blots
• -Introduced in the mid 1970s
• -Drawback limited to examining a small number
  of genes at a time.

12
Hybridization methods continued

• SAGE(Serial Analysis of Gene Expression)
• -Introduced in 1995.
• -Simultaneously measures expression levels of a
  large number of genes.
• -uses short sequence tags to mark the
  transcripts of a gene and to identify the number
  of transcripts generated by each gene.
• -Drawbacks - time consuming, involves multiple
  steps and extensive sequencing to identify
  appropriate tags.

13
Important discoveries that led to current
microarray technology

• In the late 1980s robotic devices made it
  possible to spot a surface in a compact and
  regular pattern.
• 10,000 spots on a 22 X 22 cm2 surface
• In the mid-1990s, the number of genes assayed in
  an experiment was increased by further reducing
  the pitch ( the center to center distance between
  the spots) to 200-400µm.
• Up to 2,500 spots /cm2
• Sample required was reduced
• Now can do 3,000 spots/cm2

14
One of the first experiments involving microarrays

• Quantitative Monitoring of Gene Expression
  Patterns With a Complementary DNA Microarray.
  (Schena et al. 1995)
• Compared expression levels of genes in wild-type
  Arabidopsis thaliana and a HAT4 (transcription
  factor) transgenic line of A. thaliana.
• Transgenic involves the transfer of foreign DNA
  into cells.
• Chose A. thaliana because it has the smallest
  known genome of any higher eukaryote.

15
Schena article continued

• Results showed a 50-fold elevation of HAT4 mRNA
  in the transgenic line compared to the wildtype.
• Expression of all other genes differed by less
  than a factor of 5 between the HAT4-transgenic
  and wild type plants.
• Showed that the gene transfer of the HAT4
  transcription factor to A. thaliana was
  successful and the HAT4 gene caused phenotype
  changes in the transgenic line such as earlier
  production of flowers, altered pigmentation, and
  poor germination.

16
A second experiment done in the Schena article

• Compared the expression levels of genes in cells
  from the root and cells from the leaf.
• A comparison of the scan revealed widespread
  differences in gene expression between root and
  leaf tissue.
• mRNA from the light regulated CABI gene was 500
  times more abundant in the leaf.
• The expression of 26 other genes differed by a
  factor of 5 between the leaf and the root.

17
A recent major discovery involving the
sub-classification of lung cancer cells

• Molecular Profiling of Non-Small Cell Lung Cancer
  (NSCLC) and Correlation With Disease-Free
  Survival (Wigle et al. June 2002)
• Idea - To use gene expression to develop
  molecular classifications of cancers and
  correlate gene-expression with disease-free
  survival.

18
Methods of the experiment

• Samples were collected from tumor cells of 39
  patients with NSCLC and frozen in liquid nitrogen
  until used.
• Patients were given the same treatment for lung
  cancer.
• After treatment, patients were monitored with a
  minimum of a 1 year follow up, to see if the
  cancer recurred or not.

19
Methods continued

• 24 patients experienced relapse of their tumor.
• 15 patients remained disease-free based on both
  clinical and radiological testing.
• Previously frozen tumor cells from patients who
  relapsed were labeled with 1 dye and tumor cells
  from patients who didnt relapse were labeled
  with another dye and the gene-expression levels
  were compared using microarrays.

20
(No Transcript)
21
Results/Conclusions

• The patients were clustered hierarchically on the
  basis of 2899 genes.
• Two groups emerged that appeared to separate
  patients that relapsed compared to those that
  remained disease free.
• This experiment provides evidence that relapse
  risk for NSCLC can be determined from gene
  expression data from cDNA microarrays.
• This could lead to improvements in prognosis and
  patient management for patients with NSCLC.

22
(No Transcript)
23
(No Transcript)
24
cDNA Microarray
25
Agenda

• Procedure and Technology
• Image Analysis
• Data Processing and Analysis
• Experimental Design
• Advantages and Disadvantages

26
Procedure Overview
27
Procedure Microarray Fabrication

• Select probes from EST databases such as GenBank.
• cDNA probes cloned in bacteria
• Amplify by PCR, purify the product (0.6 to 2.4
  kb)
• Poly-L-Lysine Coating on the glass slide
• Slides coated with poly-L-lysine have a surface
  that is both hydrophobic and positively charged.
  The hydrophobic character of the surface
  minimizes spreading of the printed spots, and the
  charge appears to help position the DNA on the
  surface in a way that makes cross-linking more
  efficient.
• Printing
• DNA cross-linked to the substrate by ultraviolet
  radiation.
• Slide Blocking - Succinic anhydride reduce
  positive charge, prevent target hybridizing with
  substrate

28
Printing Robot Albert
29
Printing Technologies

• Photolithography
• Mechanical
• Microspotting
• Inkjets

30
Procedure cDNA Targets Preparation

• Extract RNA from tissue, using oligo-dT primers.
• Reverse Transcription and Flourescent Label the
  cDNA samples using Cy3-dUTP or Cy5-dUTP
• Hybridization to the slide

31
Procedure Hybridization

• Far more probes in each spot on array than there
  are targets that can hybridize to them, so no
  competition between target
• Hybridization chambers is used to avoid
  evaporation

32
Electron Micrograph picture
33
Scanning Technology

• Confocal Laser
• Sensitive, high resolution
• Stimulate Cy3 and Cy5 one at a time or
  simultaneously using a filter.
• CCD camera-based (digital camera)
• Cheaper, viable with many different floures,
  collect signal over long time, but less
  sensitive, low resolution

34
Confocal Laser Scan Principle
35
Image analysis

• Gridding
• Segmentation
• Intensity Extration
• Background correction
• Target detection

36
Gridding

• Assign coordinates to the spots
• Can be automatic or manual
• Issues Rotation/Skew of Array and overall
  shift

37
Segmentation

• Define individual spot boundaries
• Types of Methods
• Fixed Circle
• Adaptive Circle
• Adaptive Shape
• Historgram

38
Segmentation Continued

• Fixed Circle
• impose a boundary of constant diameter.
• problem not all signals are circular and same
  size.
• Adaptive Circle
• Estimate the diameter for each spot.
• problem signals are not all circular.

39
Segmentation Continued

• Adaptive Shape
• Different size and shapes possible
• Seeded Region Growing assign a seed region, and
  compare value with neighbor regions, and use
  algorithm to merge regions.
• Historgram
• Read in values over a larger area then the spot
  and use histogram to plot the values
• Assign an adaptive threshold to determine whether
  a pixel belongs in the foreground or background.

40
Intensity Extraction

• Foreground and Background Intensity Extracted.
• After background correction, Target Intensity is
  derived by subtracting background intensity from
  foreground intensity.

41
Background Correction

• Why? spot intensity may contain contribution
  from contaminations, non-specific binding.
• Methods
• Local Background
• Morphological Opening
• Delineate large area around the spot, remove all
  spots, and estimate background.
• Constant Background

42
Methods
Local Background
Morphological Opening
43
Effect of Background Correction
44
Quality Measure

• Spot size or shape
• Ratio of background intensity to foreground
  intensity

45
Data Processing

• Use the ratio of Cy3 and Cy5, log transform to
  produce normalized data.
• MA Plot log2 R/G vs (1/2) log2 (R/G)
• Normalization
• Correct systematic errors of dyes, pins and
  background.
• Use set of house-keeping genes or all genes as
  non-changing control
• Filtering disregard extreme measurement values.
  Example throw out top 5 and bottom 5 of
  target intensities measurements.

46
Normalization Examples
Normalization for print-tip group
locations Differently colored lines represent
different print-tip groups.
Yellow control genes Cyan mixed sample pool
titration Red best fit curve for entire sample
Before
After
47
Data Analysis

• Design database to hold the data and perform
  analysis.
• Data Mining -- the automated extraction of hidden
  predictive information from databases
• Supervised use of predefined class. support
  vector machines
• Unsupervised hierarchical clustering, k-mean
  cluster, self-organizing maps.

48
Experiment Design

• Direct vs. Indirect
• Variance differs by factor of 4
• Direct more precise because comparison is within
  slides
• Indirect more feasible in some circumstances
  (comparsion between 3 samples)
• Dye Swap reduce systematic difference between
  dyes
• Replications
• Take multiple readings of same experiment
• Repeat experiment (new RNA extraction)

49
Advantages and Disadvantages

• Advantages
• Flexible probes can be custom made
• Higher specificity due to longer probes
• Simultaneous hybridizations in comparative
  studies (minimize experimental variation)
• Disadvantages
• Large amount of RNA required.
• Cross Hybridization Risk (false positives)
• Probes Multiple Constraints with Glass Possible
• Relatively fewer features than oligo approach

50
Oligonucleotide Microarrays

• One Gene Representation on GeneChip
• The Photolithographic Construction of Microarrays
• Major Steps
• Intensity Calculation
• Data Format
• Advantage Disadvantage

51
Overview Affymetrix GeneChip Images  
52
Overview Affymetrix GeneChip Images  
53
Overview GeneChip Expression Analysis Process
54
Overview Catalog GeneChip Expression Arrays
55
Gene Representation
56
Photolithographic Construction
57
GeneChip Single Feature
A single feature on an Affymetrix GeneChip
microarray. 
58
GeneChip Hybridization
GeneChip Hybridization of tagged probes to
Affymetrix GeneChip microarray. 
59
Hybridized GeneChip Microarray
Hybridized GeneChip Microarray Cartoon
depicting scanning of tagged and un-tagged
probes on an Affymetrix GeneChip microarray. 
60
Major Steps
61
(No Transcript)
62
Intensity Calculation
Each probe cell 10x10 pixels. Gridding
estimate location of probe cell centers.
Signal  Remove outer 36 pixels 8x8
pixels.     The probe cell signal, PM or MM,
is the 75- th percentile of the 8x8 pixel
values. Background Average of the lowest 2
probe cell values is taken as the background
value and subtracted.
63
Affymetrix Microarray Suite (MAS) 5.0

• Experiment Information Files (.EXP)
• - containing information on array type,
  sample information, fluidics settings and
  hybridization scanner settings.
• Image data Files (.DAT) 
• - the raw image file direct from the scanner
  with the analysis grid.
• Cell Intensity Files (.CEL)
• - containing measured intensities locations
  for an array that has been hybridized.

64
Affymetrix Microarray Suite (MAS) 5.0

• Analysis output File (.CHP)
• - the output generated by the analysis of a
  .DAT or .CEL file. 
• Report File (.RPT)
• - summarizes background noise, housekeeping
  information spiked-in controls
• Chip Description Files (.CDF).
• - contains names and locations for each gene
  represented on the chip.

65
Advantages

• High specificity
• Can use small amount of RNA
• Widely used, so annotation of probe sets is of
  relatively high quality.

66
Disadvantages

• Very expensive to design. (US300,000)
• Expensive to perform experiments. (US400 300
  labeling/hybridization)
• Limited to the species for which there are chips
  available sequence required.
• Single target hybridization, so comparison always
  involves two experiments, and dye swaps are
  impossible.

67
The future of microarrays

• Search for a higher density whole-genome chip
  that could simultaneously measure the expression
  of all 30,000-40,000 human genes.
• If they are eventually made, they wont be cheap
  and there could be problems with storage and
  analysis.
• Shift from chips made by research labs to chips
  manufactured by companies.
• Companies will make standard gene chips for
  different organisms and also custom chips.

68
Future of microarrays continued

• Fluorescent labeling to electrical detection
• Fluorescent labeling interferes with
  hybridization because of steric hindrance, and
  requires very expensive detection systems.
• Ideally we could quantify the amount of
  hybridization by measuring an electrical signal
  and not having to modify the sample before
  hybridization.
• Measuring an electrical signal could be done by
  monitoring changes in electrical properties upon
  hybridization or by weighing the extra mass of
  hybridized material.

69
Future of microarrays continued

• Better interpretation of data.
• Development of better software for clustering and
  correlation.
• Make a standard data format.
• Expression data could be stored in a database.

70
Limitations of Microarray Technology

• Does mRNA expression profile accurately reflect
  protein abudance in cells?
• Gygi Article correlation of mRNA and protein
  abundance in yeast only .356
• Protein Microarrays

71
References

• Campbell, N. A., Reece, J. B. (2002). Biology.
  San Fransisco Benjamin Cummings.
• Schena, M. et al. (1995). Quantitative Monitoring
  of Gene Expression Patterns with A Complementary
  DNA Microarray. Science, 270(5235), 467-470.
• Sebastiani, P. et al. (2003). Statistical
  Challenges in Functional Genomics. Statistical
  Science, 18(1), 33-70.

72
References continued

• Strachan, T., Read, A. P.(1999). Human Molecular
  Genetics. New York A John Wiley Sons, Inc.
• Wigle, D. et al. (2002). Molecular Profiling of
  Non-small Cell Lung Cancer and Correlation with
  Disease-free Survival. Cancer Research, 62,
  3005-3008.
• Website
• http//www.dna-arrays.com/index.html

73
References continued

• Lectures
• Deonier, R., Microarray Analysis. 10/21/2002
• Calabrese, P., Microarrays/Gene Expression
  Arrays. 3/11/2003