README.md

---
license: mit
tags:
- biology
- protein-classification
- microalgae
- genomics
- nanoGPT
- metagenomics
- tara-oceans
pipeline_tag: text-classification
---

# algaGPT

Causal language model for binary classification of microalgal vs. contaminant protein sequences.

## Model Description

- **Architecture:** nanoGPT (Andrej Karpathy)
- **Task:** Binary classification of microalgal protein sequences via next-token prediction
- **Mode:** TI-inclusive (full-length sequences)
- **Training data:** ~58.6M protein sequences (1:1 algal:contaminant ratio)
- **Algal sources:** 166 microalgal genomes across 10 phyla
- **Contaminant sources:** Bacterial, archaeal, and fungal sequences from NCBI nr

## Performance

| Metric | Score |
|--------|-------|
| Recall | >99% |
| Speed vs. BLASTp | ~10,701x faster |

## Usage

**Input:** Protein sequence (amino acid string)
**Output:** Classification tag (algal/contaminant) via next-token prediction

## Applications

algaGPT was used as the primary proteome extraction tool in the ELF-NET study (Nelson et al., forthcoming), where it purified algal protein sequences from 2,044 TARA Oceans metagenome assemblies, yielding 221.9 million sequences for downstream domain-environment coupling analysis.

## Authors

David R. Nelson, Ashish Kumar Jaiswal, Noha Samir Ismail, Alexandra Mystikou, Kourosh Salehi-Ashtiani

Green Genomics Lab, New York University Abu Dhabi

## Citation

```bibtex
@article{la4sr2025,
  title={Pan-microalgal dark proteome mapping via interpretable deep learning and synthetic chimeras},
  author={Nelson, David R. and Jaiswal, Ashish Kumar and Ismail, Noha Samir and Mystikou, Alexandra and Salehi-Ashtiani, Kourosh},
  journal={Patterns},
  volume={6},
  pages={101373},
  year={2025},
  publisher={Cell Press},
  doi={10.1016/j.patter.2025.101373}
}
```

## Contact

Kourosh Salehi-Ashtiani — ksa3@nyu.edu