Protein Function Prediction Pipeline

This project provides a pipeline for protein function prediction using deep learning and protein language model embeddings.

Quick Start

1. Prepare Data:

   • Use prepare_data.py to process your raw FASTA and TSV files into training-ready data.

2. Extract Embeddings:

   • Generate embeddings for your protein sequences (see plm.py or cluster_embed/).

3. Train Model:

   • Run the main training pipeline:

     python train_script.py

   • The script uses configs in configs/ and saves results in outputs/ or runs/.

Structure

• train_script.py: Main entry for training
• prepare_data.py: Data preparation
• configs/: Experiment/model configs
• Network/, models/: Model code
• utils/: Utilities