Skip to content

ahm3draxa/CRPTS

 
 

Repository files navigation

CRPTS

CRPTS: Predicting Transcription Factor Binding Sites using DNA Shape Features Based on Shared Hybrid Deep Learning Architecture

Tested on

Setting up

Clone the repositopry into your working space.

Data preparation

Download the data from https://bitbucket.org/wenxiu/sequence-shape/get/2159e4ef25be.zip Firstly, using encode.sh script to preprocess DNA sequences and their corresponding shape features.

  • Usage: bash encode.sh
  • 'pbmdata' denotes the path of storing experimental data, e.g. /yourpath/pbmdata.

Run CRPTS using DNA shape and DNA sequences or CRPT using DNA sequences

  • Usage: you can excute run.sh script directly, in which you should modify the python command accordingly, e.g.: python train_val_test_hybrid.py -datadir ./pbmdata/$eachTF/data -run 'shape' -model 'CRPTS' -batchsize 300 -k 5 -params 30 --train
  • The command '-run' means 'shape' using four shape features, and the command '-model' can be a choice of {'CRPTS', 'CRPT'}
  • Note that you should change the ouput path in the run.sh script, the naming rule is: 'model_' + args.model + '_' + args.run.
  • Type the following for details on other optional arguments: python train_val_test_hybrid.py -h

Contact

About

Updated with Python3 Support

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 98.0%
  • Shell 2.0%