ccMPRA data, first run bins_to_fasta.sh then train_test_split.sh
gkm-svm.sh to train a gkm svm and test model performance, gkm-kernels.sh to get the precomputed SVM kernels.
gkmmatrix.c contains modified code of the lsgkm package, allowing to calculate a matrix for the test data as well.
combined_kernel_SVM.py constructs an SVM using a combination of precomputed promoter and CRE kernels and tests this on the test set.