Metadata-Version: 2.4
Name: respepbind
Version: 0.0.1
Summary: ResPepBind: A deep learning predictor of peptide-protein binding interactions
Author-email: Arslan Siraj <arslan.qau11@gmail.com>, Sabir Ali <Sabirkhokhar110@gmail.com>
Maintainer-email: Arslan Siraj <arslan.qau11@gmail.com>, Sabir Ali <Sabirkhokhar110@gmail.com>
License-Expression: MIT
Project-URL: Repository, https://github.com/Arslan-Siraj/ResPepBind
Project-URL: Documentation, https://github.com/Arslan-Siraj/ResPepBind#readme
Project-URL: Issues, https://github.com/Arslan-Siraj/ResPepBind/issues
Project-URL: Weights, https://github.com/Arslan-Siraj/ResPepBind/releases/tag/ResPepBind_model_v1.0.0_seed2026_10fold
Keywords: peptide-protein interactions,peptide binding,protein language models,ProtT5,deep learning,bioinformatics
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Science/Research
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Operating System :: OS Independent
Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: torch<3,>=2.2
Requires-Dist: numpy<2,>=1.24
Requires-Dist: scikit-learn<2,>=1.3
Requires-Dist: transformers<4.50,>=4.38
Requires-Dist: sentencepiece>=0.1.99
Requires-Dist: safetensors>=0.4
Requires-Dist: protobuf<6,>=4.25
Dynamic: license-file

# ResPepBind

A 10-fold residual neural network ensemble for peptide-protein interaction prediction.

The official checkpoint ZIP is hosted separately as a [GitHub Release](https://github.com/Arslan-Siraj/ResPepBind/releases/tag/ResPepBind_model_v1.0.0_seed2026_10fold). **The first prediction automatically downloads all 10 fold checkpoints** into `~/.cache/respepbind/models/`. It also downloads the pretrained ProtT5 encoder from Hugging Face when missing residue embeddings are first generated. 

## Predict: input FASTA or CSV (CPU or GPU)

Install package with `pip install respepbind` from PyPI published distribution.

FASTA:

```text
>positive_1
Peptide: MLGP
Protein: MAEVPPGPSSLLPPPAPPAPAAVE

>negative_1
Peptide: ACDW
Protein: MLEVKQRL
```

CSV:

```csv
id,peptide,protein,true_label
pair1,MLGP,MAEVPPGPSSLLPPPAPPAPAAVE,1
pair2,ACDW,MLEVKQRL,
```

Run:

```bash
respepbind --fasta examples/pairs.fasta --device cpu --output-csv results/fasta.csv
respepbind --csv examples/pairs.csv --device cuda:0 --output-csv results/csv_gpu.csv
```

Or use wrappers:

```bash
./scripts/predict_cpu.sh --fasta examples/pairs.fasta --output-csv results/cpu.csv
CUDA_VISIBLE_DEVICES=0 ./scripts/predict_gpu.sh --csv examples/pairs.csv --output-csv results/gpu.csv
```



