[R]MicroArray Preprocessing

Preprocessing includes
– Image analysis and data import
– Background adjustment
– Normalization
– Summarization (for Affymetrix)
– Quality assessment

Practise Data:

使用limmaGUI匯入two-channel microarray data

參考: http://bioinf.wehi.edu.au/limmaGUI/

A. 下載 Swirl Zebrafish資料

http://bioinf.wehi.edu.au/limmaGUI/Swirl/swirl.zip

B. 安裝 Bioconductor

> source(“http://bioconductor.org/biocLite.R”)

> biocLite()

C. 安裝以下packages: convert, limma, arrayQuality, marray, mclust, hexbin, limmaGUI, sma, tkrplot, R2HTML, statmod

1. 程式套件 > 選擇存放處 > CRAN, CRAN(extra), BioC software
2. 程式套件 > 安裝程式套件 (或用install.packages(“convert”)

利用affylmGUI匯入two-channel microarray data:

準備動作:

A 安裝estrogen package以取得estrogene dataset:

1. 程式套件 > 選擇存放處 > BioC experiment
2. 程式套件 > 安裝程式套件

B. 安裝affylmGUI packages:

1. 程式套件 > 選擇存放處 > BioC Software
2. 程式套件 > 安裝程式套件

Normalization

To identify and remove systematic sources of variation in the measured fluorescence intensities, such as

– different labelling efficiencies of the dyes;

– different scanning parameters;

– sector/print-tip, spatial, or plate effects, etc.

不要做過多的normalization, 因為可能把原本是signal的data弄不見.

Self-self hybridization: 去探討dye的系統性誤差的典型例子; 使用相同的sample, 不同的染劑, 去做染色就可以知道不同染劑的差異性, 理論上呈現的pattern應該是如下左圖, 但實際上會是長得像右圖的:

所以從以上的例子來看, 矯正(Normalization)的動作變得很重要.

可以用: Within-Array Normalization.

實作:

使用limmaGUI來作Normalization:

Normalization > Select Within-Array Normalization