Normalization and Data Processing

All expression data was obtained from Affymetrix MOE 430 2.0 chips. Normalization and expression modeling were performed in the R statistical computing environment, with extensive use of the Bioconductor suite of R packages. In particular, probe level summaries were computed using the GC-RMA method (Wu et. al (2003)). The GC-RMA normalized data for each treatment and for the entire study are available for download through this site.

Purification Schemes

This link leads to a table showing the schemes used to purify each cell type. All cells were primary cells freshly isolated using a combination of magnetic and flow cytometric sorting with the indicated markers. Initial populations were derived from the tissue indicated in the third column (1)- activated T cells were obtained by stimulation with conconavalin A for 8-11 hours in vitro. Because of the difficulty in obtaining RNA from terminally differentitated erythroid cells, the erythrocyte population is the only one that contains substantial numbers of cell-type specific progenitors. The correlation coefficient (R2) that indicates the similarity between the two arrays for each population is indicated in the leftmost column.

Correlation Coefficients

Shown below is a table of the correlation coefficients between each microarray chip. An R2 of 0.96 or greater was seen between replicates.

Normalized Data

The link below contains an Excel worksheet of the normalized data set.

CEL Files

Each link below leads to the raw CEL file for the chip indicated.

Note: .gz files are about 6.2MB and .bz2 files are about 4.6MB.