Lecture
Computational Epigenetics
Max Planck Institute for Molecular Genetics
Bioinformatics
Bioinformatics!
Excurs:
Clustering, very popular „red green pictures“
How do i get these pictures?
What can i read from them?
Are they „stable“?
Lecture: Computational Epigenetics
Distances
Given Vectors x = (x1, …, xn), y = (y1, …, yn)
n
• Euclidian Distance: d E ( x, y ) = ∑ ( xi − yi ) 2
i =1
n
• Manhattan Distance: d M ( x, y ) = ∑x −y .
i =1
i i
• Correlation Distance:
∑ ( x − x )( y
i i − y)
d C ( x, y ) = 1 − i =1
.
∑ ( xi − x ) 2 ∑ ( yi − y ) 2
i =1 i =1
Lecture: Computational
Vorlesung: Microarray Datenanalyse Epigenetics Kapitel 2
Example:
Day 1 Day 2 Day 3
Gene 1 0 1 2
Gene 2 12 11 10
Gene 3 5 6 7
Gene 4 7 6 7
See blackboard
Lecture: Computational
Vorlesung: Microarray Datenanalyse Epigenetics Kapitel 2
HEP
Lecture: Computational Epigenetics
Summary:
•Chromosomes 6, 20, 22
•1.9 million CpGs (map to 873 genes)
•12 tissues
•Differential methylation ~ evolutionary conservation
•17% differentially methylated in 5‘UTR -1/3 anticorrelation with transcription
•CpG density ~ methylation
Lecture: Computational Epigenetics
Type and distribution of amplicons analysed in HEP
Lecture: Computational Epigenetics
Example region
Lecture: Computational Epigenetics
Correlation of DNA methylation with spatial distance
Lecture: Computational Epigenetics
Correlation of DNA methylation with cell type
Lecture: Computational Epigenetics
DNA methylation in relation to distance to TSS
Lecture: Computational Epigenetics
DNA methylation, age, sex
CD4+ lympho
Age Sex ELK1
vs.
(X chr)
Dermal fibro
Control
Sex specific
Lecture: Computational Epigenetics
DNA methylation and evolutionary conserved regions
Lecture: Computational Epigenetics
DNA methylation and transcription status
Lecture: Computational Epigenetics
Conservation of DNA methylation between human and mouse
Lecture: Computational Epigenetics
Bioinformatics!
These databases and datasets therein provide a solid
basis for further bioinformatic explorations
Lecture: Computational Epigenetics
Schedule
Lecture 1+2: 06.01.2009 Introduction; Databases
Lecture 3: 12.01.2009 Highthroughput technologies,
genomewide DNA methylation profiling
Lecture 4: 13.01.2009 Analysing multivariate genomic data
Lecture 5: 19.01.2009 DNA methylation prediction
Lecture 6: 20.01.2009 Evolutionary aspects
Lecture 7: 26.01.2009 Prediction of imprinted genes
Lecture 8: 27.01.2009 Deamination followed by Substitution
Klausur II: 28.01.2009
Lecture: Computational Epigenetics I
Highthroughput Technologies
Lecture: Computational Epigenetics
HIGHTHROUGHPUT:
Technology
Sequencing RNA:
Hybridisation RNA: -RNAseq
-Macroarray -(SAGE)
-cDNA
-Oligoarray Interaction:
-Illumina -Chipseq
Interaction: DNA
-ChIPchip -(Sequencing)
DNA
-arrayCGH
-SNParray
Max Planck Institute for Molecular Genetics
ChIP chip
Max Planck Institute for Molecular Genetics
ChIPseq
Max Planck Institute for Molecular Genetics
ChIPseq platforms
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Max Planck Institute for Molecular Genetics
Specific Issues for processing the data
Lecture: Computational Epigenetics
DNA methylation profiling
1. Weber et al, 2005, 2007 Nature meDIP
2. Rakyan et al, 2008 Genome Research mPOD
3. Meissner et al, 2008 Nature
4. Yagi et al, 2008 Genome Research
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Microarray Hybridization
tissue 1 total RNA 1
total RNA 2 tissue 2
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Comparison of human, chimpanzee and rhesus
sequence was used to define CpG loss and CpG gain in the human lineage
Lecture: Computational Epigenetics
Comparison of promoter methylation profiles of germline-specific genes versus total
genes in WI38 primary fibroblasts and sperm cells. The density plots show that most
germline-specific genes are hypermethylated in somatic cells and unmethylated in
sperm. Only ICPs and HCPs are considered.
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
methylation profiles of DNA (mPod)
Problem of MeDIP:
- Interpretation of binding signal not transformable into absolute methylation
signal:
-impossible to estimate absolute methylation levels fromMeDIP
experiments
- analysis of CpG-poor regions, in particular, has been assumed to be
difficult (almost completely methylated but signal low)
Solution: Batman = Bayesian tool for methylation
analysis
Lecture: Computational Epigenetics
BATMAN:
- Transformation of normalized MeDIP-chip log2-ratios into a quantitative
measure of DNA methylation across a wide range of CpG densities
- assumption: methylation only at CpG
- Adjustment results in a quantifiable measure of methylation across (all) CpG
density values
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Bisulfite treatment and ChIPseq
Not DNA methylation sensitive
... C - - C G G ...
... G G C - - C ...
... C C G C G G ... Fill in
... G G C G C C ...
Lecture: Computational Epigenetics
extend
C G G AAAAAAAAAAAA
G C C TTTTTTTTTTTTT
Amplify and ChIP seq
Lecture: Computational Epigenetics
1. DNA methylation patterns are better correlated with histone methylation
patterns than with the underlying genome sequence context.
2. methylation of CpGs are dynamic epigenetic marks that undergo extensive
changes during cellular differentiation, particularly in regulatory regions
outside of core promoters.
3. analysis of embryonic-stem-cell-derived and primary cells reveals that ‘weak’
CpG islands associated with a specific set of developmentally regulated
genes undergo aberrant hypermethylation during extended proliferation in
vitro, in a pattern reminiscent of that reported in some primary tumours.
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
Lecture: Computational Epigenetics
-Similar procedure as Meissner et al (different enzymes)
-Identification by use of mouse promoter tiling array
Lecture: Computational Epigenetics
MAT SCORES
Lecture: Computational Epigenetics