Final Project of Team 1

1
Final Project of Team 1

• To build a statistical tool web-site for advanced
  Micro-Array data analysis

???? ??? Masuya ??? ??? ???
2
Version context

• This version of power-point is written for
  bio-informatics course HW5s guide.
• And will be taken as a part of TEAM 1s final
  project output.

3
Document Outline

• A.Introduction
• A.1)Purpose
• A.2)Process flow
• A.3)Our website
• A.4)Resources
• B.Process flow
• B.1)Statistic Significant Test -SAM
• B.2)Datasets preparation and adjustment
• B.3)Clustering data

4
A.Introduction
5
A.1.Purpose

• Perform statistics analysis to get the
  discriminator markers. (Question 1)
• Cluster analysis to distinguish tumor from
  non-tumor. (Question 2)
• GO analysis of discriminator markers.

6
Introduction of Micro-Array

• An array is an orderly arrangement of samples. It
  provides a medium for matching known and unknown
  DNA samples based on base-pairing rules and
  automating the process of identifying the
  unknowns.
• The sample spot sizes in microarray are typically
  less than 200 microns in diameter and these
  arrays usually contains thousands of spots.
  Microarrays require specialized robotics and
  imaging equipment that generally are not
  commercially available as a complete system.

7
Introduction of Micro-Array(cont.)

• Types of microarray
• Format I probe cDNA (5005,000 bases long) is
  immobilized to a solid surface such as glass
  using robot spotting and exposed to a set of
  targets either separately or in a mixture. This
  method, "traditionally" called DNA microarray, is
  widely considered as developed at Stanford
  University. A recent article by R. Ekins and F.W.
  Chu (Microarrays their origins and applications.
  Trends in Biotechnology, 1999, 17, 217-218) seems
  to provide some generally forgotten facts.

8
Introduction of Micro-Array(cont.)

• Format II an array of oligonucleotide (2080-mer
  oligos) or peptide nucleic acid (PNA) probes is
  synthesized either in situ (on-chip) or by
  conventional synthesis followed by on-chip
  immobilization. The array is exposed to labeled
  sample DNA, hybridized, and the
  identity/abundance of complementary sequences are
  determined. This method, "historically" called
  DNA chips, was developed at Affymetrix, Inc. ,
  which sells its photolithographically fabricated
  products under the GeneChip trademark. Many
  companies are manufacturing oligonucleotide based
  chips using alternative in-situ synthesis or
  depositioning technologies.

9
A.2.Micro-Array data analysis process flow
10
Micro-Array data analysis process flow
(Statistic sub-flow)
11
A.3.Our Website

• http//linhots.no-ip.com/biap

What you can do From it?

• You can download all dataset from our web-site.
• You can download all the power-points from our
  web-site.
• You can find all statistics background
  information from our website.
• You can do some data adjustments on our website.

12
A.4.Resource

• Shi, L. 1998. DNA Microarray ( Genome Chip ) --
  Monitoring the Genome on a Chip.
  http//www.gene-chips.com/
• Leung, Y. F. and D. Cavalieri. 2003.
  Fundamentals of cDNA microarray data analysis.
  Trends Genet. 19 649 - 59.

13
Process flowB.1.Significant test
14
Our approach

• We use SAM as significant process tool.
• It is a client-side application embedded in
  office Excel
• You can download all the information on our
  website (include SAM.zip software).
• Or you can go to http//www-stat.stanford.edu/tib
  s/SAM/index.html to download SAM.zip.

15
Is Your Raw data Normalized?

• SAM cannot perform normalization
• So if the data is not normalized
• Feed it to the clustering tool
• In the Adjust data tab, normalize it
• Save the data
• Now the data is ready to be fed into SAM

16
Where
http//rana.lbl.gov/

• We can download executable software, source code,
  user manual from this website

17
1.Data Preparation
The input data file may also look like the one
shown below. Yellow regions in the diagram show
that they are optional.

• A minimal cluster input data would
• look like the one shown below

18

• Cluster will inform you about the loaded data

19

• Here normalization is done

20
1
21

• Note
• For the latest version of the clustering tool,
  after pressing Apply button, the data should be
  saved using
• File-gtsave as-gt filename

22
Data is Normalized!

• The new file saved contains normalized data.
• It can easily be worked on the SAM tool.

23
System Requirements
SAM

• Approach using Internet Explorer
• latest Microsoft Java Virtual Machine if run with
  XP.
• The Microsoft Data Access Components.
• Microsoft Excel 97 or higher.
• If your XP still can not install or run SAM, use
  win2000 or win98.

24
Installation.3-1
SAM

• Download the file after registration from
  http//www-stat.stanford.edu/tibs/SAM/index.html
• After registration you will get a email with
  password in it. Then go to

25
Installation.3-2
SAM
After decompression, Double click on Setup.
26
Installation.3-3
SAM

• Fire up Excel and click on the Tools menu. Choose
  Addins and click on Browse. Select the directory
  where the setup process installed SAM (C\Program
  Files\SAMVB, if you chose the defaults) and click
  on the ??/?? Addin subdirectory. Double click on
  the SAM file.The SAM addin will be loaded and the
  box against the phrase Significance Analysis for
  Microarrays will be checked.Once you click OK,
  you should now see two buttons on your Excel
  toolbar named SAM and SAM Plot Control .

27
Data source format
SAM
1. Normalization data needed 2. SAM can use data
in multiple sheets.
28
Data format transformation from HCC1648
SAM

• This data structure is HCC cells
• And normal cells, and are not paired data
• Missing data
• If there is no data in sheet 2 or 3, delete no
  data sheet first
• Selection all data
• Press SAM function
• ? Two classes, unpaired with missing data

Original
Transformed
29
SAM
Parameters input

• Choice you data type
• Check your data is logged or not
• Here we use clone ID
• Missing value will be regenerated by two methods,
  choice one and how many neighbors will be
  calculated for the missing value. if there were
  missing value SAM will generate a new data sheet
  in the final output, you can use this for
  clustering analysis
• You can generate another random seed by yourself.
• Then OK

3
4
5
30
SAM
Significant gene no. listed, when you change
delta value or fold change this will change
spontaneously
False predict number
31
you can manually input the delta-value also after
you check the list Delta table or just move the
tag
32

• The number of significant genes and FDR are
  depend on the delta and fold change which you
  chose.

33
When we choice fold change is 2 and delta value
2.65, then 102 genes will be listed as
significant changes.
34
Fold change
Entering this address you will get this gene
information Showing in next slide
35
(No Transcript)
36
Process flowB.2.Data-adjustment
37
Why we need it

• In the process flow Each softwares I/O Data
  format maybe different
• We provide 2 on-line process tools
• You can reach these 2 tools from our website

1
2
38
1.Adjust the HCC datasetSeparate the NAME column
1. Upload HCC dataset
2. Set up parameters (or use default value)
3.Click the button and get the result
39
2.Adjust the HCC datasetFilter with Statistic
result
1. Upload dataset
2. Feed with Statistic result(genelist string)
3.Click the button and get the result
40
Process flowB.3.Clustering
41
Where
http//rana.lbl.gov/

• We can download executable software, source code,
  user manual from this website

42
Analyze Process Flow
Raw data collection
Data preparation
Data cleaning
Significant test by SAM
Normalization
Data-mining
Clustering
Visualization
View the result
43
1.Data Preparation

• Format your data as follows.

Simple form
Complete form
The dif whether we want to take weight and order
into accounted
44
Import data into the software

• Cluster will inform we about the loaded data

1
2
3
45
More on filter tab
46
2.Clustering

• The tool provides us with three methods of
  clustering
• Hierarchical
• K-means
• Self organizing map ( SOM )

47
2-1.Hierarchical Clustering

• The hierarchical clustering tab allows to perform
  hierarchical clustering on the datasets.

48
2.2 Partitioned Clustering

• The parameters controlling K-Means clustering
  are
• The number of clusters ( K )
• The maximum number of cycles.

K-means may be faster!
49
2.3 SOM Clustering

• Here we have considered one dimensional approach
  of SOM

50
3.Result Viewer
We need a viewing tool
Micro-array Raw data
.CDT and .GTR files
Cluster
Tree-View
Data cleaning Data-mining
Clustering result Visualization
51
3.Result Tree-viewer

• Use the mouse click/drag/find
• Use the keyboard up/down/left/right

2
1
3
52
Appendix
53
1What Is Good Clustering?

• A good clustering method will produce high
  quality clusters with
• high intra-class similarity
• low inter-class similarity
• The quality of a clustering result depends on
  both the similarity measure used by the method
  and its implementation.
• The quality of a clustering method is also
  measured by its ability to discover some or all
  of the hidden patterns.

54
2. This is Micro-Array Data-Set
Exp 1
Exp 2
Exp 3
Exp 4
Exp 5
Exp 6
Gene 1
Gene 2
Gene 3
Gene 4
Gene 5
Gene 6
55
3. Take Each row(gene) as a line
Log2(cy5/cy3)
56
Then we can have similarity Correlation
Correlation0.97
57
4. Finally we can do cluster
58
5. Hierarchical Clustering
Hierarchical clustering. Define Similarity
Choose cluster gene? And if we calculate the
weight?
59
6. Hierarchical Clustering

• Define Similarity metric
• Center or not
• Absolute or not
• Linear or not

60
7. Hierarchical Clustering

• Center or not

• Absolute or not

61
8. Hierarchical Clustering

• Linear or not

62
9. Hierarchical Clustering

• During construction of the hierarchy, decisions
  must be made to determine which clusters should
  be joined.
• The distance or similarity between clusters must
  be calculated.
• The rules that govern this calculation are
  linkage methods.

63
10.. Partition Clustering

• K-mean, Professor had taught us.
• K-Mean/K-Medoid
• K-Medoid choose a new node which is more near the
  central instead of central point.

64
WE SEPARATE them!!!!!
65
11. Ordering of operations

• 1.log transform of all the values
• 2.Mean centre rows
• 3.Median centre rows.
• 4.Normalize rows
• 5.Mean centre columns
• 6.Median centre columns
• 7.Normalize columns