Array comparative genomic hybridization (aCGH) can detect copy number variations (CNVs) across whole genomes in a single experiment. ISACGH software allows visualization of aCGH and gene expression array data on chromosomal coordinates to identify regions of altered copy number and correlate these with expression levels. It provides four methods to estimate genomic copy number from aCGH data and allows functional annotation of altered regions. The software generates publication-quality figures and allows exploration of copy number alterations across multiple samples.

1. Gene gain and loss: aCGH
ISA CGH
IV Course on Microarray data analysis
March 13, 2008. Valencia
Rafael C. Jimenez

2. Genetic Variation in human genomes
From chromosome anomalies to single nucleotide changes
Chromosome Gene Nucleotide

3. CNV/CNP
Copy number variation of DNA segments
 Deletions
 Insertions
 Duplications
Mb
b(nt)
Kb

4. Comparative Genomic Hybridization

Method for analyzing genomic DNA for unbalanced genetic alterations (CNV)

Based on fluorescent hybridization of DNA
Techniques

FISH, PCR, Southern

Array CGH

Clones (BAC, YAC, PAC)

PCR non-redundant

Oligonucleotides

cDNA

5. Example: Array CGH
BAC probes
 Clones to cover a genomic region
 Extraction and purification
 Array DNA onto glass slides
 Hybridization of labeled normal and tumor genomic DNA to the microarray
 Analyze fluorescence ratio

6. Example: Array CGH
BAC probes

7. Array CGH
Advantage

Analysis of whole-genome in a single experiment

Higher resolution than conventional CGH (5-10 Kb)

Inability to detect mosaicism, balanced chromosomal
translocations, inversions and whole-genome ploidy changes

Depending on the plataform chosen:
- cDNA: limited by the genes encoded on chromosomes,
cross-hybridizations
- BACs, PACs: DNA amplifications are necessary
Limitations:

8. Array CGH

9. ISA CGH
ISACGH allows visualizing array
CGH data or/and expression
arrays onto human or mouse
chromosomal coordinates
(automatically found through their
standard identifiers) and represents
the copy number alterations found
by using different methods.
Correlations between copy number
and gene expression level can
easily be observed in a plot. The
program allows finding minimal
common regions with altered copy
number across different arrays.
http://isacgh.bioinfo.cipf.es

10. General outline
INPUT: gene expression or/and
genomic hybridization values.
OUTPUT: prediction of regions with
alterations in the number of copies,
plotted in the same figure than gene
expression values to view the
relationship between both variables
∙Four methods for the estimation of
genomic copy number
∙Other parameters:
- Scale
- Management of multiple probes
http://isacgh.bioinfo.cipf.es

11. Probe identifiers
Different array platforms for CGH
measurements (BACs, cDNA clones
and oligonucleotides for array spots)
Users must collect and keep updated the
chromosomal coordinates of the probes
ISACGH automatically retrieves them
and plots the hybridization values over
the corresponding positions in
chromosomes.
Different IDs can be uploaded
(Ensembl Ids, accession, Unigene,
HUGO, RefSeq, Affy, BAC names...)
User-defined information on
chromosomal coordinates can
alternatively be supplied to the program
Chr1:203465273-20348412

12. Genomic copy number estimation
Smoothing: variation of the Adaptive Weights Smoothing method implemented in the program GLAD.
Binary segmentation: checks whether every single point in the data set is a breakpoint in an iterative way. It uses a
permutation distribution to test for mean differences between groups drawing robust inferences that do not rely
upon any model of the data
Regression: uses some characteristics of linear regression of intensity measure on position along the genome.
Such a regression line has slope close to zero when fitted in regions with homogeneous intensity levels but the slope
differs form zero when intensity measures on the left-hand side are higher or lower than intensity measures in
the right-hand side. Given the points of intensity measurements ordered by their position in the
chromosome, our procedure fits a regression line for each N consecutive points. Thus a vector of slopes equivalent in
some way to a derived curve of the original data is obtained.
t-test is used to to identify the slope-peaks that are big enough to indicate a break-point in the intensity levels.
Isowindow: tries to identify borders between regions with a significant change in the values of intensity of
hybridization. Given the intensity measurement points from the array ordered by their position in the chromosome, a
first step finds those that are good candidates of being such borders. Roughly speaking, a point will be a
good candidate if the p-value of a t-test comparing some close points located at its left and right neighborhoods
is low enough.
The binary segmentation method uses the global distribution of the dataset while the others are more based on the
local distributions of the points.

13. Plotting data and copy number estimation
One array / All chromosomes Expression data
Breakpoint estimation

14. Plotting data and copy number estimation
One chromosome / All arrays

15. Studying breakpoints and details at gene level
Click to get a detailed representation of the
chromosomal region spanning 4Mb that
includes the corresponding probes of the
array located therein as well as genes
mapped by Ensembl. A blue line represents
the estimation of the genomic copy number.

16. Minimal common regions with consistent copy
number alterations across several arrays
ISACHG estimates the genomic copy number in each individual arrays and then merge
this information in a unique plot.
The plot represents those regions that are either consistently gained or lost in
all the arrays. The average hybridisation intensity is also represented.

17. Correlation between genomic and expression data
Quite useful when studying the effect of genomic copy number
alterations on the expression level of genes is to have a simple
representation of the relationship between both variables.

18. Functional annotation of altered copy number regions
Functional annotation of chromosomal regions based on the gene ontology (GO) annotations.
Annotations for regions of gain or loss are compared to the background of annotations
corresponding to the rest of genes present in the array, in order to see whether this
region has any characteristic functionality or not.
A Fisher exact test for contingency tables is performed to look for GO terms significantly
overrepresented in the chromosomal region studied compared to the background abundance of
GO terms in the rest of the genome.
P-values adjusted for multiple testing using the FDR method.

19. An example

20. Ejemplo
1. Choose an
organism

21. Ejemplo
2. CGH array data

22. Ejemplo
3. Class labels (optional)

23. Ejemplo
4. Gender labels (optional)

24. Ejemplo
5. Choose the copy number estimation mehod
Fast
Precission
· Isowindow
· Binary
Segmentation
·Regression
· Smoothing

25. Ejemplo
6. Gene expression data

26. Ejemplo
7. Position data (optional)

27. Ejemplo
8. Scale value (optional)

28. Ejemplo
8. Scale value (opcional)
Scale: default
(max)
Scale: 5

29. Ejemplo
9. Reference lines (optional)

30. Ejemplo
9. Reference lines (optional)

31. Ejemplo
10. If you have replicates
· Average
· Max
· Min

32. Ejemplo
10. Define tipo de representación

33. Ejemplo
11. Choose the representation

34. Ejemplo
12. You can download map data

35. Ejemplo
13. Download data form regions

36. Ejemplo
14. Check correlation expression-copy number

37. Ejemplo
15. Zoom y functional analysis
ISA zoom

38. Ejemplo
15. Zoom y functional analysis
Ensembl DAS zoom

39. Ejemplo
15. Zoom y functional analysis
Análisis funcional
(FatiGO)

40. Ejemplo
15. All arrays plotting

41. http://gepas.bioinfo.cipf.es/cgi-bin/tutoXX?c=/isacgh/isacgh.config

42. Acknowledgments
Joaquin Dopazo

Group leader
Lucia Conde

ISACGH developer

Tutorial

Slides
Ignacio Medina

Support, guidance and advice
David Montaner

Support, guidance and advice
Francisco Garcia

Support, guidance and advice