README: for NCI 60 data
=======

We would like to thank Dr. Jane Fridlyand (from UCSF) for providing 
us with the processed NCI 60 data. Please refer to Dudoit et al. 2002
(JASA 97: 77 -- 87) for details of pre-processing. The paper is also
available as a technical report at:
http://www.stat.berkeley.edu/~sandrine/tecrep/576.pdf

*The following files are in the correct format for our Java code.

We randomly partitioned the original data with 61 samples into 
a training set with 43 samples and a test set with 18 samples.

Training set: train_NCI60_5244_43_1.txt
- consists of 5244 genes and 43 experiments, spanning 8 different
  cancer cell lines.
- The rows represent the genes.
- Each column represent an experiment. The type of cell line is
  indicated in the header row.
- Unfortunately, the gene names are not available from the pre-processed data, 
  so we made up unique identifiers for the names (ie. Gene_XX).
- Note that repeated measurements are not available (in general) for this
  data. So, each entry represents a log expression ratio.

Training set: test_NCI60_5244_18_1.txt
- consists of 5244 genes and 18 experiments, spanning 8 different
  cancer cell lines.
- in a similar format as the training set

Classes: NCI60_8class.txt
- 8 classes
- This file specifies the name of each of the 8 classes, and is required 
  by our ewusc software.

Label for each experiment in the training set: NCI60_label43.txt
- total 43 experiments
- This file specifies the label for each experiment, and is required by
  our ewusc software.

Label for each experiment in the test set: NCI60_label18.txt
- total 18 experiments
- This file specifies the label for each experiment, and is required by
  our ewusc software.