/ Home / Analysis Tools / Normalization

DNA microarray resultst normalization

Version 2.1

• The R language
• Launching the “Goulphar.R” R script
• Script execution
• Graphical outputs created
• Normalised data
• Useful links

This tutorial could be downloaded as a PDF file (1.1 Mb). The older version of this tutorial using Varan is available using this link.

The GenePix software enabled you to convert your microarray image pixels to an intensity digital data for each spot. The next step consists in normalizing data within each experiment to eliminate systematic errors and scaling all of them so that they can be compared.

The R language:

R is a statistical and graphic programming language close to the S+ language. It is available freely on the Windows, MacOS and Linux platform. R functions could be improved and are based on the contributions of a developer community. R has a growing room in biology, a large number of modules have been developed to answer specifically high-throughput biology problems, including the microarray domain. So, an independent R project, Bioconductor has been dedicated to microarray.

R modules installation is quite simple, particularly under Windows operating system. An executable installer installs R automatically and modules are loaded from the graphic interface using the item “Packages” from the user interface menu bar.

In order to normalize our microarray data, we will use a R script developed locally, which allow to call a set of R functions and to obtain a file of normalised data and a lot of graphical outputs to control each step of the normalisation process. More information is available on the script web pages.

Launching the “Goulphar.R” R script:

1. Open this web page link. You will find there the R script to download “Goulphar.R”. Copy this script in the folder where you already save the GPR (GenePix Result) file of your slide. Be careful to use the “Save link target as…” function when you right-click on the script link.
   
   
   
2. Create the “param_goulphar.dat” file using the form available on the website. The various options available are explained in the script online user manual. Be careful, as previously, to use the “Save link target as…” function when you save the parameter file.
   
   
   
   
   
   
   Be careful, if you want to edit the parameter file, using Excel for example, you must save the file using a tabulated text format and not the Excel native format.
   
   
3. Launch the R program on your computer displayed the user interface. In the working window, a prompt begin the line (>), this means that R is waiting for your commands.
   
   
4. You must go to the folder where your GPR, “Goulphar.R” script and parameters files are located. To do so select in the menu bar, the function “Change dir…” and choose using the window browser the folder where your files are stored.
   
   
   
5. Verify using the “getwd()” command, that you are located in the expected folder. Be careful, under the Windows environment, file paths are like this “C:\your_path\your_folder\». R does not understand anti-slash “\”, they are replaced by slashes “/”.
   
   
   
6. Enter the command “source(“Goulphar.R”)” to launch the script. It works using as inputs your GPR file and the normalisation parameters found in the file “param_goulphar.dat”.
   
   
   

Script execution:

1. R loads the modules called by the “Goulphar.R” script.
2. Next it loads the parameters from the file “param_goulphar.dat”, they will be displayed on screen.
3. Next R load the GPR file then it:
   1. filters data;
   2. normalises;
   3. builds graphical outputs;
   4. and creates normalised data output file.
4. 4. Here is what you can observe in the R command line interface:
   
   
   Be careful, if you launch the script several times with various parameters or on other GPR files, from the same working directory, all results files previously created will be erased and replaced without any alert messages!
   
   
5. 5. Quit R using the “q()” command. It is not necessary to the save the workspace as it creates huge files.
   
   

Graphical outputs created:

Details on each graphics created are available on the Goulphar online user manual.

Filtered spots depend of the choices made in the parameter file. They combine spots excluded because of their smaller diameter, saturating intensities or GenePix flags. They will be excluded form the normalisation process and from most of the graphical outputs. They will be all there instead, if not any filter is selected.

Normalised data:

R script graphical outputs are automatically stored in the working directory, the one from which the “Goulphar.R” script is launched. A tabulated text file is created at the end of the script (“GPR_start_file_norm.txt”). The R, G, Rb and Gb columns correspond to raw foreground and background intensities in Cy5 and Cy3 respectively. They are not modified by the normalisation process, as the A column (corresponding to the log2 mean intensities). The Mnorm, alone, includes normalized data. The Rb and Gb columns, background measurements in Cy5 and Cy3, are found only if the subtracted background option has been selected in the parameter file. Cells labelled “NA” are cells where values are none available, discarded after the filtration process.

Useful links:

• Goulphar has been created to allow the quick normalisation of microarray data and the visula monitoring of this step. It offers a user form to help normalisation parameter selection. Website: http://transcriptome.ens.fr/goulphar/