This archive contains the Level 3 data of  
212 HNSC primary solid tumor samples and 50 cell line control samples.
The cell line samples are not related to TCGA, but they are printed on all the 
slides (including those with TCGA samples) to serve as a control
for further normalization of the data (e.g. combing two batches).

Level 3 data are derived from Level 2 data. The normalization is described
as follows:

1) Starting with a level 2 data matrix, median center the data of each protein;
2) From 1), calculate the median value for each sample;
3) subtract the sample medians from 2) from the level 2 data of the corresponding samples.

This method has proved to work well. However, the savvy user is encouraged try other 
normalization methods if she likes.


A word about batch effect correction: 

There may be two or more batches for a certain cancer type, e.g. BRCA, UCEC, etc.
We are working on methods of batch effect alleviation that is 
specific to this platform. currently, if there is more than one batch,
the batches are normalized separately and uploaded as is.







