Webbased Microarray Tools

1
Web-based Microarray Tools
Team 3

• Chia-Chi Yeh (R93922048)
• Tzu-Kuo Huang (R93922002)
• Yu-Feng Huang (D92922019)
• Tai-Wei Wu (R93842007)

2
Outline

• Motivation
• Features
• System Overview
• Built-in Functions
• Data Flow
• Demonstration

3
Motivation

• Every analysis tool has its own user interface -gt
  a long learning phase.
• Each tool only provides few functionalities -gt
  transfer from one tool to another frequently.
• Some tools are commercial -gt no way to customize
  them to fit ones needs.

4
Features

• Friendly user interface We provide a uniform
  interface for every function.
• Flexibility It is easy to import modules in our
  system.
• Transparency All the details of underlying
  packages are transparent to the end-users.
• Open source Everything we used is publicly
  available on the internet.

5
System Overview
intensity
intro
modules
dataset
plot
function
t-test
process
submit
browser
pca
status
wrapper
CEL file
hcluster
result
6
Built-in Functions

• Plot a probe intensity image for an Affymetrix
  CEL file. intensity
• Compare gene expression levels between two
  samples. plot
• Determine genes that have significantly different
  expression levels in two types of samples.
  t-test
• Use PCA(Principle Component Analysis) to perform
  clustering. pca
• Hierarchical Clustering. hcluster

7
Data Flow (1/2)

• From users view
• Read introduction of this website. intro
• Choose a function to perform. function
• Upload files and submit the job. submit
• Watch job status and wait. status
• During waiting, a user may choose another
  function and submit jobs.
• Get the result when a job is done. result

8
Data Flow (2/2)

• From systems view
• Show available functions. function
• For a chosen function, render the input fields
  and submit form. submit
• Check function, options, and uploaded files of a
  submitted job. process
• If no error, execute the corresponding module at
  the background. wrapper
• Monitor all the submitted jobs. status
• Show the result when a job is done. result

9
Demonstration

• Now we are going to demonstrate how to use our
  system to complete homework 5.
• We use the HCC1648 dataset, and the input files
  are available on our website.
• Please read this guide carefully and make sure
  the input files are correct, or you will waste
  everyones time.

10
Step 1 Prepare Files (1/3)

• The input file is compatible with the
  TAB-delimited text file in Microsoft Excel.
• The first row contains sample names, the first
  column contains gene names, and other cells are
  real numbers.
• Our system will simply replace missing values by
  zero. If you want to handle them in a different
  way, please impute them manually before uploading.

11
Step 1 Prepare Files (2/3)

• Download HCC1648 dataset.
• Open it with Excel.
• Remove useless rows and columns.
• Edit the gene names if you want.
• Put normal samples (Liver) in one sheet, and
  tumor samples (HCC) in another.
• Save them as TAB-delimited text file.

12
Step 1 Prepare Files (3/3)
Sample names
Impute the missing value or system will fill it
by zero.
Save as TAB-delimited text file
Gene names
13
Step 2 Select a Function (1/2)
Click!
14
Step 2 Select a Function (2/2)
For example, click t-test.
15
Step 3 Upload Files
16
Step 4 Submit a Job
Click!
17
Step 5 Get Results (1/3)
Wait until it is done.
18
Step 5 Get Results (2/3)
Step 5 Get Results (2/3)
Click to view the results.
19
Step 5 Get Results (3/3)
Step 5 Get Results (3/3)
20
Useful Information

• You can submit many jobs at the same time, and
  the status page shows every job you submitted in
  this session.
• The results of each function are described in the
  result page.
• If you close your browser, your session and
  submitted jobs are lost.

21
About Homework 5

• First run t-test and get the results.
• Then delete entries in your input files which are
  not in gene.txt.
• Run one of the clustering functions using the
  cleaned input files.
• If no errors, the results are what you want.
• If you have problem with hcluster, just use pca
  or kmeans instead.