SoluBat: A Bidirectional Mamba Framework for High-Throughput Protein Solubility Prediction in Bioprocess Optimization

Introduction (SoluBat)

SoluBat is a sophisticated hybrid model designed for precise protein solubility prediction, incorporating the Mamba model.

Results

Paper Results

Features

• Bidirectional Mamba model boosts accuracy in protein solubility prediction.
• Dynamic gating integrates multimodal features with high efficiency.
• Near-linear complexity reduces GPU usage compared to Transformers.
• Built-in residue-level attribution enhances biological interpretability.
• Extensive benchmarking confirms strong generalization and industrial applicability.

Requirement

Please make sure you have installed Anaconda3 or Miniconda3.

conda env create -f environment.yaml
conda activate SoluBat

Configuration

Parameter	Description
data_root	Root directory for raw data
num_folds	Number of cross-validation splits
fold_idx	Fold index (-1 for all, 0-6 for single)
batch_size	Training batch size
lr	Initial learning rate
max_epochs	Maximum training epochs

Usage

python scripts/train.py

Contributing

Contributions and suggestions from the community are welcome! If you find a bug or have an improvement suggestion, please submit an issue or a pull request.

License

This project is licensed under the MIT License.