The original data were from http://www.fruitfly.org/seq_tools/datasets/Human/GENIE_96/splicesets/
The data partition for 6-fold cross-validation were based on Martin Reese's partitions, 
whose gene list given in the file "combined_GB.sets" (http://www.fruitfly.org/seq_tools/datasets/Human/GENIE_96/combined_GB.sets). 

The training and testing datasets for 6-validation are in 6 sub-folders (cv0 - cv5).
Each sub-folder contains the following 3 files:

	1) DonorTrT.mf   -> Training (True sites) 
	2) DonorTrF.fa   -> Training (False sites) 	
	3) DonorTest.fa  -> Testing 

***Note***
Training and testing datasets in "cv0" sub-folder were directly converted from the original data 
 at http://www.fruitfly.org/seq_tools/datasets/Human/GENIE_96/splicesets/ 

Prepared by weichun huang, weichun.huang@duke.edu