//standard input format with annotation
1
// type: when type=1,we evaluate whether there is fusion between every two genes in your file
// when type=2,we evaluate whether there is fusion between  0th gene and 1th gene, 2th gene and 3th gene,(i)th gene and (i+1)th gene
gene0
//line with only one 'word'(no space inside) is regarded as gene name thus a separator for data of different gene
1 3 2 4 5
3 6 7 2 8
//one line for each exon,in which data of each experiment is seperated by space
//The gene you use shouldn't have too many exons, less than 15 for the best.
//This is due to LSTM's input size.
//If you want to do prediction on genes with much more exons,you can adjust the code, and train a model yourself.
//The code is on github:    https://github.com/faded53222/Predict-fusions-using-exon-counts
4 2 19 0.5 0
gene1
14 53 45 67 123
23 54 63 13 424
//the number of exons can differ in different genes
//the number of experiments should be same
gene2
90 123 435 542 123