10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006....
Transcript of 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006....
![Page 1: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/1.jpg)
Microarrays
10-810: Advanced Algorithms and Models for Computational Biology
![Page 2: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/2.jpg)
Why sequence is not enoughIdentifying genes and control regions is not enough to decipher the inner workings of the cell:• We need to determine the function of genes.• We would like to determine which genes are activated in which cells and under which conditions.• We would like to know the relationships between genes (protein-DNA, protein-protein interactions etc.).•We would like to model the various dynamic systems in the cell
![Page 3: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/3.jpg)
Central Dogma
DNA
mRNA
transcription
translation
Proteins
Transcription factors
![Page 4: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/4.jpg)
Genes and Gene ExpressionTechnologyDisplay of Expression Information
![Page 5: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/5.jpg)
Genomic DNA
Promoter Protein coding sequence Terminator
What is a gene?
![Page 6: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/6.jpg)
TFIIB
RNAPII
TT
T T
M
T
A
How are Genes Regulated?DNA-binding Activators Are Key To Specific Gene Expression
G
Cc
Ta
![Page 7: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/7.jpg)
T
R
TT
T T
M
T
A
How are Genes Regulated?DNA-binding activators are key, but there are additional factors
G
Cc
Ta a
rcctcRRR
![Page 8: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/8.jpg)
Activators
Genome-wide Gene Expression (mRNA) can be Measured with DNA Microarrays
GeneRNAPIITFIIH
Transcription apparatus
mRNA
mRNA label hybridization ATGC
TACG
![Page 9: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/9.jpg)
![Page 10: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/10.jpg)
Some Additional Numbers
Yeast:6200 genes~200 transcriptional regulators
Human:30,000 - 50,000 genes~1700 transcriptional regulators>100 cell types
![Page 11: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/11.jpg)
Genes and Gene ExpressionTechnologyDisplay of Expression Information
![Page 12: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/12.jpg)
Microarray Hybridization• Watson-Crick base pairing of complementary DNA
sequences.
• Microarrays have thousands of spots, each representing a piece of one gene, immobilized on a glass slide.
• The intensity (or intensity ratio) of each spot indicates the amount of labeled cDNA hybridized, thus, representing the starting mRNA transcript abundance.
![Page 13: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/13.jpg)
Two major technologies• cDNA arrays
- probes are placed on the slides- allows comparison of different cell types
• Oligonucleotide arrays- partial sequences are printed on the array- measure values in one tissue type
![Page 14: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/14.jpg)
Hybridization and Scanning— cDNA arrays
- Prepare Cy3, Cy5-labeled ss cDNA
- Hybridize 600 ng of labeled ss cDNA toglass slide array
- Scan
![Page 15: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/15.jpg)
Cartesian PixSys 5500 withquill printing technology
• Complete subsequences are printed on the array•10,000 spots/slide• Spots are 100-200 µm in diameter• Hybridization volumes: 20-100ul
![Page 16: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/16.jpg)
Array Scanning
Laser based - fluorescent emission
![Page 17: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/17.jpg)
Hybridization and Scanning—oligo arrays
![Page 18: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/18.jpg)
cDNA vs. Oligo: Pros and Cons
cDNA
• Does not require sequence
• Cheap
• Direct comparisons
• Inaccurate
• Cannot measure individual samples
Oligo
• Can be designed to minimize cross hybridization
• Allows for internal control
• Both lead to better accuracy
• expensive
• limited to certain species
![Page 19: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/19.jpg)
Errors
Microarrays introduce many errors which should be taken into account when working with measured expression values:• Scanning errors• Spotting errors• Cross hybridization• Errors related to day / reading device / experimentalist• Background differences between slides
![Page 20: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/20.jpg)
Error typesMicroarrays introduce many types of errors which should be taken into account when working with measured expression values:• Scanning errors additive + multiplicative• Spotting errors multiplicative• Cross hybridization multiplicative• Errors related to day / reading device / experimentalistadditive + multiplicative• Background differences between slides additive
![Page 21: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/21.jpg)
Handling the Different Errors• Scanning errors• Spotting errors• Cross hybridization• Errors related to day / reading device / experimentalist• Background differences between slides
Analysis of image data (we assume it was performed)
![Page 22: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/22.jpg)
Handling the Different Errors• Scanning errors• Spotting errors• Cross hybridization• Errors related to day / reading device / experimentalist• Background differences between slides
Use ratio instead of individual values:
Yi = Ri / Gi
![Page 23: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/23.jpg)
Handling the Different Errors• Scanning errors• Spotting errors• Cross hybridization• Errors related to day / reading device / experimentalist• Background differences between slides
For Oligo arrays, use the match / mismatch spots
![Page 24: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/24.jpg)
Match / Mismatch
• Presence and absent calls can be made using the Match / Mismatch information.
• However, it has been reported that in some cases the mismatch was higher than the match.
![Page 25: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/25.jpg)
Handling the Different Errors• Scanning errors• Spotting errors• Cross hybridization• Errors related to day / reading device / experimentalist• Background differences between slides
Normalization (next lecture)
![Page 26: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/26.jpg)
Binding arrays
• Instead of printing the genes on the microarray, we can print the intergenic region (an area upstream of the gene).
• We tag a protein of interest (a transcription factor) and fuse all proteins to DNA.
• Next, we hybridize the extracted portions of DNA onto the array, resulting in areas that are bound by the TF being spotted on the microarray.
![Page 27: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/27.jpg)
Genes and Gene ExpressionTechnologyDisplay of Expression Information
![Page 28: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/28.jpg)
Yeast cell cycle
expression program
genes
Experiments (over time)
100 20 70 80
gene 1
Higherexpression compared to baseline
Lower expression compared to baseline
baseline expression
Spellman et al Mol. Biol. Cell 1998
![Page 29: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/29.jpg)
alph
a 0’
alph
a 7’
alph
a 14
’al
pha
21’
alph
a 28
’al
pha
35’
alph
a 42
’al
pha
49’
alph
a 56
’al
pha
63’
alph
a 70
’al
pha
77’
alph
a 84
’al
pha
91’
alph
a 98
’al
pha
105’
alph
a 11
2’al
pha
119’
/Ndd1
Mcm1
1 3.5 >5
1/3 1 3
Yeast Cell Cycle Gene Expression Program
Spellman et al. and Cho et al., 1998
800 Genes
![Page 30: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/30.jpg)
>9 >6 >3 1:1 >3 >6 >9Fold repression Fold induction
amming in Response to Oxidant0 10 20 40 60 120
minutes
enome expression is rogrammed
6218 genes
ROS response
![Page 31: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/31.jpg)
600 Conditions/Mutations
6200
Gen
es
Single-gene MutationsEnvironment
Exercising the Genome
![Page 32: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/32.jpg)
Visualization:Relative vs. absolute expression
![Page 33: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/33.jpg)
Using annotation databases• Statistical tests to identify
the overlap with various functional categories
![Page 34: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/34.jpg)
![Page 35: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/35.jpg)
Genome wide binding
genes
experiments (transcription factors)
TF1 TF8
gene 1
Probably bound by this TF
no binding by this TF
Lee et al Science 2002
![Page 36: 10-810: Advanced Algorithms and Models for Computational …epxing/Class/10810-06/new7.pdf · 2006. 11. 27. · alpha 0’ alpha 7’ alpha 14’ alpha 21’ alpha 28’ alpha 35’](https://reader036.fdocuments.us/reader036/viewer/2022071217/604a2eef5404e266ba70a58c/html5/thumbnails/36.jpg)
What you should know• The basic idea behind microarray profiling• The two different microarray technologies• Pros and cons for each• Noise factors in microarray experiments (more next time)