Normalisation

  restart analysis

help  

Background

This tool was developed under the ActinoGEN consortium funded by the European Commission FP6. The aim of this tool is to support high-throughput analysis of gene expression in Streptomyces coelicolor using arrays distributed and developed under the consortium (see our microarray site ). Thus, this normalisation tool follows the optimal workflow of a Streptomyces microarray-based experiment created through our own experience (see our guidelines/workflow ). If you should require a particular analysis method please contact us at microarrays@surrey.ac.uk.

Please note that this software utilises R (http://www.r-project.org/) and the R package Limma (Smyth (2005). Limma: linear models for microarray data.)

Data uploaded and generated on this site will be kept for 24 hours only, please make sure you download your results within this time-frame.

Working through the workflow

1. File entry

The files currently accepted for uploading and normalisation are the raw output files generated by the feature extraction software BlueFuse (files ending with _output.xls) and Agilent Feature Extraction (files ending with .txt); files that do not follow these formats will cause errors to be produced.

We recommend BlueFuse for feature extraction of our spotted PCR- and Oligo-based microarrays and Agilent Feature Extraction for our high-density ink-jet in situ synthesised (IJISS) arrays.

Multiple files of a suitable format (BlueFuse or Agilent Feature Extraction) can be selected for uploading. Each file has to be selected individually and can be removed (by clicking on the remove button) from the file upload list if you make a mistake. Once all files for normalisation have been selected, click on the upload button to upload your files. Feedback for the upload process will be given.

We recommend that all feature extraction output files for a single complete experiment be loaded at the same time.

2. Experimental details entry

  1. Select the feature extraction software used.
  2. Select the experimental design from which your output files are from.
Click on the Submit button

3. Spatial effect check and within-array normalisation selection

Here you will be able to visualise the reconstructed images of your arrays using raw data. Please note that in these images the colours may not relate directly to the label-dye of your samples (Green (Cy3) or Red (Cy5)). If you have uploaded Agilent Feature Extraction data red in the image represents the sample loaded in your gProcessedSignal channel and the green in the image represents the sample loaded in your rProcessedSignal channel. If you have uploaded BlueFuse data red in the image represents the sample loaded in your AMPCH1channel and the green in the image represents the sample loaded in your AMPCH2 channel. Please note that the image has been turned such that the top-left of the array is the bottom-left of the image.

The images are given as an aid in deciding the normalisation method to apply and/or to identify arrays that should not be included in the analysis. If you have uploaded more that one data file you will be able to scroll through the images using the controls above the image (< to scroll back, > to scroll forward and X to close the image panel (n.b. once closed you cannot re-open this panel)).

Once you have checked your arrays for spatial effects you can choose the within-array normalisation method for your experiment. Note that the normalisation method choices given are those that are recommended in our guidelines for your selected experimental design e.g. Loess for cDNA vs cDNA or Global Median for cDNA vs gDNA.

Normalisation methods:

For more information please see: Smyth, G. K., and Speed, T. P. (2003). Normalization of cDNA microarray data. Methods 31, 265–273.

Once you have selected the normalisation method to apply click on Submit.

4. Within-array normalisation check and across-array normalisation selection

Here you will be able to visualise the reconstructed images of your arrays using within-array normalised data. Please note that in these images the colours may not relate directly to the label-dye of your samples (Green (Cy3) or Red (Cy5)). If you have uploaded Agilent Feature Extraction data red in the image represents the sample loaded in your gProcessedSignal channel and the green in the image represents the sample loaded in your rProcessedSignal channel. If you have uploaded BlueFuse data red in the image represents the sample loaded in your AMPCH1channel and the green in the image represents the sample loaded in your AMPCH2 channel. Please note that the image has been turned such that the top-left of the array is the bottom-left of the image.

The images are given as an aid in deciding the normalisation method to apply and/or to identify arrays that should not be included in the analysis. If you have uploaded more that one data file you will be able to scroll through the images using the controls above the image (< to scroll back, > to scroll forward and X to close the image panel (n.b. once closed you cannot re-open this panel)).

In addition to reconstructed images of arrays you will find a box-plot for all of the arrays in your uploaded experiment. Ideally each box in this plot should look similar as we assume the null hypothesis that there is no difference between the samples in your uploaded experiment. If your boxes are not uniform this would suggest that you will need to perform an across array normalisation step (which can selected in this part of the workflow).

Once you have checked your arrays and box-plots you can choose the across-array normalisation method for your experiment. Note that the normalisation method choices given are those that are recommended in our guidelines for your selected experimental design.

Normalisation methods:

For more information please see: Smyth, G. K., and Speed, T. P. (2003). Normalization of cDNA microarray data. Methods 31, 265–273.

Once you have selected the normalisation method to apply click on Submit.

5. Normalisation and processing finished, data ready for downloading

Here you will find a box-plot for all of the arrays in your uploaded experiment, to show the effects of the across-array normalisation applied. You will also find a link to a downloadable zip file containing all of the necessary files for downstream analysis:

close help

Status: START > File entry