It is important to specify the chiptype you've used, otherwise annotations based-analyses can't be performed later on.
There are three background removal methods, in addition to the option to not apply any removal. The simplest method imply subtracts the background signal form the foregound (spot) intensities (which have the potential of introducing negative values if not used in conjunction to an offset value, see below). The second possibility is to use Edward's smoothing method for background removal, which explicitly tapers the values in order to avoid negative number output. Finally, the normexp method is often found to be superior to other methods, especially when combined with background offset of 50. Do note, though, that arrays with specific anomalies, like having a few background spots with extremely high intensity level, may benefit from not having any background subtraction at all. Please refer to the Ritchie et al. paper below for more details.
Within array normalization methods offer basic options. Either no normalization is applied, arrays are translated to the same median, or a loess smoother is used for removing possible nonlinear (dye) effect from the data. Between array normalization options are no normalization, translation to the same median and quantile. It is also possible to use variance stabilizing normalization. Please note that during normalization the data is also log2-transformed
Agilent arrays contain many control probes which might interfere with the analysis. Thus, control probes are removed at user's discretion. Note however that after the control probes have been removed, you will not be able to perform pathway analysis using the gene set test -tool, since unfiltered normalized data is required for that analysis. If you would like to remove the control probes, you have to mark the ControlType column as "Annotation" when importing the data to Chipster with the Import tool.
A tab-delimited text file containing gene names, descriptions and expression estimates. This file is suitable for all further analyses. Normalization also generates a phenodata table you should fill in before any further analyses are conducted.
This tool uses Bioconductor package limma.
For normalization, please cite:
Smyth, G. K., and Speed, T. P. (2003). Normalization of cDNA microarray data. Methods 31, 265-273.
For background correction steps, please cite:
Ritchie, M. E., Silver, J., Oshlack, A., Silver, J., Holmes, M., Diyagama, D., Holloway, A., and Smyth, G. K. (2007). A comparison of background correction methods for two-colour microarrays. Bioinformatics.