Biomedical Engineering Reference
In-Depth Information
Figure 2 . Heat map of the genes differentially expressed in diffuse large B-cell lymphoma data
(DLBCL) and follicular lymphoma data (FL). Each row corresponds to a different gene,
whereas each column is a different patient. For each gene, blue indicates downregulation and
red upregulation. For gene selection we used a combination of the SNR, t -score, and
Genes@Work methods. The data used for gene selection were produced at the Whitehead
Institute (WI) (49) and are drawn to the left of the yellow line. The genes upregulated an down-
regulated in DLBCL vs. FL in the WI data show a consistent behavior in an independent data
set produced at Columbia University (CU) (52), as seen in the heat map to the right of the
yellow line.
gene expression data set produced from patients with the same diseases, but us-
ing different chips, different technicians, different patients, and different labs.
These differences are likely to break the systematic errors that can arise if ex-
periments are performed in the same lab, and the validation is therefore more
telling about the underlying biology.
On the right side of the yellow line in Figure 2 we can observe how the
genes selected using the WI data separate an independent data set, obtained in R.
Dalla-Favera's lab at Columbia University (the CU data). This independent data
set contains 14 DLBCL samples and 7 FL samples, and was previously used in
(52). We can see that the CU data reproduce the gene expression profile found
in the WI data. It is important to notice that the WI data were collected using the
Search WWH ::




Custom Search