Documentation

Getting started
Installation

DeepPI requires a 64-bit Linux system or Mac OS and Python (supported versions are Python3: 3.8 and higher) to be pre-installed on it. To obtain DeepPI, you can download the source code and compile it yourself.

  • Repository
    • Request a zip file from the author. 
    unzip DeepPI.zip
    cd DeepPI 
    python main.py --mode <run_mode> [options] --output <output_dir>
Setting up environment

Users should make sure that their local environment supports Python. For GPU usage, CUDA and PyTorch must be supported. Install CUDA and PyTorch that are compatible with the GPUs in your local environment. If your version of CUDA and PyTorch are incompatible, it will not work properly. So please check the official PyTorch documentation (https://pytorch.org/get-started/previous-versions/) to make sure. A sample installation. 

CUDA 11.0
cuDNN 8.0
Pytorch 1.7.1
Parameters

The following are arguments that the user passes to the code for computation. The parameters are passed during the execution of the program. All information about the parameters can also be found by running python main.py --help on the command line.

To run DeepPI from the command line

python main.py --mode <run_mode> [options] --output <output_dir>
  • Run mode

    • Run mode for full or partial execution. 

    --mode

    all : the entire process

    image : make image database

    filtering : filtering

    dataset : make train & test dataset (*.npy)

    model : run the deep learning model

  • Output

    • The output path.

    --output
  • Input FASTA file

    • The input is protein database file. It is required in all mode and image mode.

    --input
  • Input image database directory

    • The image database directory path. It is required in filtering mode and dataset mode.

    --image
  • Input dataset (trainset & testset) directory

    • The dataset (*.npy) directory path. It is required in model mode.

    --dataset
  • Image type

    • The type of image generator (BASIC, SNAKE, SPIRAL, HILBERT).

    --type
  • Top-N

    • Parameter N for distribution-based filtering.

    --topN
  • Threshold T

    • Parameter T for threshold-based filtering.

    --threshold
  • Epoch

    • Maximum epochs for training.

    --epoch
  • Logs

    • The log directory path.

    --logs
  • GPU

    • Specify the GPU to use.

    --gpu