The source code of paper 'Identifying modifications on DNA-bound histones with joint deep learning of multiple binding sites in DNA sequence'
npz files can be loaded as:
import numpy as np
np.load(file, allow_pickle=True)
Each data entry includes five parts, the key and the meaning of the corresponding value are as follows:
- dna: dna sequence consisting of four bases (A\T\C\G)
- dnase: dnase values corresponding to the bases in the dna sequence
- tlabel: labelling of data under the iHMnBS setting
- label: labelling of data under the DeepHistone setting
- peaks: the location of peaks in ChIP-seq data, and the encoding for histone modification that occurs in this region
You can get the model up and running simply by running the main.py file with the following command:
python main.py
For the data input, you just need to fill the first argument of the get_fn function in the main function with your data path in the main.py file.
The subsequent code in main function deals with the slicing of the data when we train the model. As the name implies, train_valid_set and test_set represent the training set and test set respectively, and later train_set and valid_set subdivide the training set into a set for training and a set for validation. This part of the code used for data segmentation can be commented out as appropriate when you test the code.
Any more questions please let me know: [email protected]