PhaSeq: predicting protein liquid-liquid phase separation with structural transfer learning

This repository contains the download link of the code for PhaSeq, a predictor of protein liquid-liquid phase separation.

PhaSeq could be download via Google Drive (faster) or our local server (slower).

How to use:

An Anaconda python environment is recommend. Check the requirments.txt file, but primarily:

• Cython>=0.28
• numpy>=1.10
• pandas>=0.15
• scikit-learn>=0.17
• xgboost==1.1.0
• torch==0.4.1
• biopython==1.78
• tqdm>=4

After entering the file directory of PhaSeq, you can use the following code to install all requirments.

pip install -r requirments.txt

If users failed to install torch==0.4.1, please try

conda install pytorch=0.4.1 -c pytorch

Example: prediction of a fasta file.

python phaseq.py uniprot.fasta uniprot.txt

Input file: uniprot.fasta

Output file: uniprot.txt

Author:

This software is written by Chunyu Yu.