This folder contains 
testfile5_hs  
train0_5_hs  
train5_hs

The human training and test set for splice sites used in G. Yeo and C.B. Burge, RECOMB 2003.

5'ss
----
Format is as follows:
> 5 
aagattg

means that 5'ss aagGTattg (i omitted the GT) is real, and 

> 0
tttaata

means that 5'ss tttGTaata is a decoy 5'ss


3'ss
----
Similarly,
> 3
tggtcccatatgaattttatt

means that 3'ss tggtcccatatgaattttAGatt is a real site (I omitted the AG), and

> 0
tggtcccatatgaattttatt

means that 3'ss tggtcccatatgaattttAGatt is a decoy 3'ss



Refer to geneyeo@mit.edu for questions regarding the dataset.