Rosaceae

Microarray analysis workshop

Time schedule: 8:30 - 10:30am, 11-12:30am (3.5 hours)

---

TODO - Split and linkify article

Workflow

1. Introduction to Microarray analysis (10mins - Marcus Davy)
2. Normalization (10mins - Robert Schaffer)
3. Bioconductor/R framework (15 mins - Marcus Davy
4. R Lab Tutorial (45mins)
5. Linear models for Microarray analysis (limma) (15 mins - Marcus Davy)
6. Experimental analysis ("45mins")

---

Bioconductor/R framework

What is Bioconductor? (http://www.bioconductor.org)

• Bioconductor is an open source development software project
• Provides tools for analysis and comprehension of genomic data
• Extensively for Affymetrix and cDNA microarray technologies
• The project started in the autumn of 2001
• Includes 23 core collaborating developers

Bioconductor Goals

The broad goals of the project are:

• To enable sound and powerful statistical analyses in genomics
• To provide a computing platform that allows the rapid design and deployment of high-quality software
• To develop a computing environment for both biologists and statisticians
• Promote high-quality dynamic documentation and reproducible research
• Using L^AT_EX, the Sweave system and tcl/tk to deliver interactive step by step pdf tutorials, e.g.

library(tkWidgets)
vExplorer()

Object oriented class method design

• Organized approach to handling large amounts of experimental data
• Class structure encapsulates the data required for microarray analysis → object
• Allows efficient representation and manipulation (including subsetting) of data from many microarray slides in an experiment
• A method is a function that performs an action on data (objects) throughout analysis

Advantages

• Newest cutting edge statistical methods available
• Modern programming language
• Powerful graphical tools available
• Its freely available

Disadvantages

• Steep learning curve
• Need to have experience programming in the R programming environment (http://www.r-project.org)
• Like all software there are bugs

Accessing Bioconductor

• Bioconductor tools are accessed using the R programming language
• R is a programming environment for statistical computing and graphics
• Initially written by Robert Gentleman and Ross Ihaka (Auckland University)
• Download R from a Comprehensive R archive network (CRAN) mirror (http://cran.stat.auckland.ac.nz)
• Install R (available for Unix, Windows, and Mac OS X)
• R version 2.2.1 has been released on 2005-12-20
• R is the environment used to design and distribute software:
  • Locally downloaded files
  • Via the internet e.g. Commands in R

# Setting proxy variable # Downloading installation script

getBioC()                                                  # Running script

Installing vs loading packages

• Packages only need to be installed once onto a computer
• Packages must be loaded with each new R session
• The R function library is used to load packages e.g.

library(limma)     #Installs the limma package 

Documentation and Help

• R manuals and tutorials are available from the R website or on-line in an R session
• R on-line help system, detailed on-line documentation, available in text, HTML, PDF, and L^AT_EX formats.

help.start()           # Browser based help documentation
help()                 # Help on a topic
? ls                   # alternative help method on ls function
apropos(mean)          # Find Objects by (Partial) Name
example(mean)          # Run an Examples Section from the Online Help
demo()                 # Demonstrations of R Functionality
demo(graphics)         # Demonstration or graphics Functionality

+R tasklist

• • tasks available from web which utilze example data available in R - object assignment, subsetting, plotting, mathematical functions, sorting etc (20 tasks?)
• Usage/interaction within environment
• Bioconductor resources/vignettes(including downloading)
• Bioconductor basics (any resources for limma out there?)

Brief about Limma (10-15 mins)

• Essentially t-statistics for each spot/gene
• Uses between gene information in moderated t-statistics
• Computationally fast/robust
• Handles missing information/use defined flag information
• benefits/limitations?
• FDR control? → ranking better than selecting cutoff

Analysis script (1+hours)

+Workshop.R

---

Scratch pad

• A flow diagram for analysis
• Recap of cDNA microarrays (slide 3)
• Microarray data issues (slide 4)
• Microarray data issues (continued)
  • Large amount of data (GPR/JPEG file size)
  • Subjective
  • Need a log of what was done so someone else can quickly reroduce the results
  • → Reproducible research (someone else can understand/reproduce the results) (McGintys talk)
• Analysis process

• R resources/contributed guides (including downloading)
• R tutorial of basics (objects/indexing/functions)