A course project of Computational Biology at Tsinghua University in the Spring semester of 2016

• Keras
• Theano or Tensorflow (as Keras backend)
• scikit-learn
• biopython

Place these raw data files in raw_data/:

• hg19.fa
• hg19_Alu.bed
• hsa_hg19_Rybak2015.bed
• all_exons.bed

Then, run

python preprocess.py

and clean data will be generated in clean_data/

python model.py -v VERIFICATION_GROUP [--alu] [--debug]

where VERIFICATION_GROUP is the index of the group used for verification, and should be an integer between 0-9.

Note that switching between using and not using Alu information (--alu flag on and off) causes the model to recompile, which may take up to several minutes.

Parameters of network and training are in config.ini.