|
This is the project page for the CELFileNormalization project. Affymetrix GeneChip data can be normalized and processed using any number of methods. We believe a tool would be useful that pre-processes the CEL file (raw data) before any of these techniques are used. The approach is simple - at the summarized image level, all chips should be adjusted to have the same general intensities.
This project consists of a java Swing application that reads in a set of CEL files, performs normalization, then writes the resulting CEL files back out. The two types of normalization that are possible include a quantile normalization (such as used in RMA, except for all possible probes on the chip) or a mean normalization (which sets the mean value of intensities on the chip as equal).
Since this program rewrites CEL files it currently works best using the old format (V3) text-based CEL files. Future work includes the read/writing of binary CEL files.
Note this project depends on jAffy for compiling. However, jar files are also available to run directly.
A zip file of all necessary jar files is available.
You can also run it via webstart.
Documentation
Background
This program was written in support of a NCI Directors' Challenge grant looking at the molecular fingerprints of disease. More specifically, a collaborative group of researchers were faced with the challenge of normalizing and processing almost 500 GeneChip arrays (HG-U133A). As a precursor to this processing, we decided to use quantile normalization of the array against a common reference chip. Then any of the processing methods (MAS5.0, RMA, DCHIP) could be used. This software is a result of this need.
Andrew Hoerter,
Steven Eschrich
|