Inference of DNA-binding preferences from HT-SELEX data using deep neural networks

The flags for the command line interface:

1. • learning_file_list: A list of HT-SELEX files. Should be written as follows: -lfl demo_data/ALX4_TGTGTC20NGA_W_0.fastq demo_data/ALX4_TGTGTC20NGA_W_1.fastq demo_data/ALX4_TGTGTC20NGA_W_2.fastq demo_data/ALX4_TGTGTC20NGA_W_3.fastq demo_data/ALX4_TGTGTC20NGA_W_4.fastq
2. • primary_selex_sequence: the sequence which is the HT-SELEX experiment primary sequence. If the selex file is of the form: ALX4_TGTGTC20NGA_W_0.fastq, the primary sequence is: TGTGTC20NGA this sequence should be supplied in the cmd. Should be written as follows: -pss TGTGTC20NGA
3. • prediction_file: Prediction data file. Should be written as follows: -pf demo_data/Alx4_1744.1_deBruijn.txt or any other predicted file
4. • output_file_location: The output file name and location. Should be written as follows: -ofl results.csv
5. • saved_model_location: If supply, saves the model in the supplied address. Should be written as follows: -sml output_model.h5
6. • loaded_model_location: Loads the model from the supplied address Should be written as follows: -lml loaded_model_name.h5

1. Training command line example:
   python deep_selex.py -lfl demo_data/ALX4_TGTGTC20NGA_W_0.fastq demo_data/ALX4_TGTGTC20NGA_W_1.fastq demo_data/ALX4_TGTGTC20NGA_W_2.fastq demo_data/ALX4_TGTGTC20NGA_W_3.fastq demo_data/ALX4_TGTGTC20NGA_W_4.fastq -pss TGTGTC20NGA -sml output_model.h5
2. Training and predicting command line example:
   python deep_selex.py -lfl demo_data/ALX4_TGTGTC20NGA_W_0.fastq demo_data/ALX4_TGTGTC20NGA_W_1.fastq demo_data/ALX4_TGTGTC20NGA_W_2.fastq demo_data/ALX4_TGTGTC20NGA_W_3.fastq demo_data/ALX4_TGTGTC20NGA_W_4.fastq -pss TGTGTC20NGA -pf demo_data/Alx4_1744.1_deBruijn.txt -ofl results.csv -sml output_model.h5
3. Using pre-trained model command line example:
   python deep_selex.py -lml test_model.h5 -pf demo_data/Alx4_1744.1_deBruijn.txt -ofl results.csv

Linux based operating system (the trained models at the "models" directory can sometime have errors under other operating systems)

python interpreter > = 3.6

python software packages:
xlrd >= 1.2.0
numpy >= 1.17.5
pandas >= 0.25.3
tensorflow version == 1.14.0
keras version == 2.3.1\