to run install the dependencies and then run:
python train.py FACTOR_NAME
e.g.: python train.py CTCF

The required dependencies are:
numpy
scipy
h5py
pybedtools
pysam
pandas
bw-python (https://github.com/dpryan79/libBigWig) 
pyDNAbinding (https://github.com/nboley/pyDNAbinding)

Additionally, directories containing the input data need to be placed into 4 
directories which must be specified at the top of baseline.py. These directories 
are:

DNASE_IDR_PEAKS_BASE_DIR
Stores the conservative DNASE peaks. 
 
DNASE_FOLD_COV_DIR
Stores DNASE fold coverage files in bigwig format. 

TRAIN_TSV_BASE_DIR
gzipped tsv files containing training regions and labels. Here's the top of an 
example file: 
chr     start   stop    A549    GM12878 H1-hESC HCT116  HeLa-S3 HepG2   K562    SK-N-SH
chr10   600     800     U       U       U       U       U       U       U       U
chr10   650     850     U       U       U       U       U       U       U       U
chr10   700     900     U       U       U       U       U       U       U       U
chr10   750     950     U       U       U       U       U       U       U       U
chr10   800     1000    U       U       U       U       U       U       U       U


TEST_TSV_BASE_DIR
gzipped tsv files containing regions to predict on. Note that: 
THE LABELS MUST BE SET BUT THEY ARE NOT USED
it requires labels because I wanted to re-use a data structure. Here's the top 
of an example file:
chr     start   stop    K562
chr1    600     800     A
chr1    650     850     A
chr1    700     900     A
chr1    750     950     A
chr1    800     1000    A