DeepSP

DeepSP: A Deep Learning Framework for Spatial Proteomics

DeepSP.py

DeepSP.py is is a python script that requires python software and some python modules to be installed.

Install dependencies

The python modules that need to be installed are listed in requirements.txt, they can be installed via the command line interface as follows

$ pip install -r requirements.txt

If any python modules fail to install, you can install them one by one with the following command

$ pip install pandas
$ pip install numpy
$ pip install sklearn
$ pip install argparse
$ pip install time
$ pip install copy
$ pip install os
$ pip install warnings
$ pip install conda install pytorch torchvision torchaudio cpuonly -c pytorch
# or 
$ pip3 install torch torchvision torchaudio

DeepSP help

The introduction of the DeepSP parameters can be achieved by naming them as follows

$ python DeepSP.py -h
usage: DeepSP.py [-h] [-f F] [-r R] [-p {Y,N}] [-n N] [-bs BS] [-s S] [-lr LR] [-l2 L2] [-eps EPS] [-nep NEP]

DeepSP for Protein Subcellular Localization Prediction

optional arguments:
  -h, --help            show this help message and exit
  -f F, --file F        file path, first column must be ProteinID, last column must be markers
  -r R, --rep R         number of repeated experiments
  -p {Y,N}, --getpred {Y,N}
                        whether to predict unknown proteins, default=Y
  -n N, --n_splits N    n-fold cross validation, default=5
  -bs BS, --batch_size BS
                        mini batch size, default=64
  -s S, -seed S         a random seed, default=0
  -lr LR, --learning_rate LR
                        learning rate, default=1e-3
  -l2 L2, --l2_regularization L2
                        l2 regularization, default=1e-4
  -eps EPS, --epochs EPS
                        training times, default=100
  -nep NEP, --nepoch NEP
                        print the results every * times, default=20

Example file

hirst2018.csv The file must be in csv format, the first column must be the protein number and there must be no duplicates, the last column must be the subcellular organelle to which the protein is localised, unknown localised proteins should be marked as unkonwn, the data in each row cannot be all zeros.

DeepSP example

DeepSP.py can be run from the command line interface with the following commands, where -f (dataset file) and -r (the number of replicates) are the two required parameters. An example command is “python DeepSP.py -f hirst2018.csv -r 3”, in which “-f” is input dataset file and “-r” is the number of replicates. There is a running example in the example.jpg image。

$ python DeepSP.py -f hirst2018.csv -r 3

wangbing587 / DeepSP