README.md

---
license:
- cc-by-4.0
size_categories:
- 1M<n<10M
tags:
- DNA Sequences
- Protein Sequences
- Computational Biology
- Bioinformatics
- Synthetic Biology
---

![image/png](https://github.com/Adibvafa/CodonTransformer/raw/main/src/banner_final.png)


# CodonTransformer Dataset

A comprehensive compilation of 1,001,197 DNA and protein sequence pairs, sourced from 164 organisms across Eukaryotes, Bacteria, and Archaea.
This dataset provides a rich resource for various computational biology and bioinformatics applications such as studying gene sequences, codon usage, and protein expression across diverse species.

## Dataset Contents

- 1,001,197 DNA-protein sequence pairs
- Sequences from 164 organisms, including:
  - Eukaryotes: Homo sapiens, Arabidopsis thaliana, Caenorhabditis elegans, Danio rerio, Drosophila melanogaster, Mus musculus, Saccharomyces cerevisiae, Chlamydomonas reinhardtii, Nicotiana tabacum
  - Bacteria: Various Enterobacteriaceae species including Escherichia coli
  - Archaea: Thermococcus barophilus, Sulfolobus solfataricus
  - Chloroplast genomes: Chlamydomonas reinhardtii, Nicotiana tabacum

## Data Collection and Preprocessing

- **Source**: NCBI resources
- **Original Format**: Gene or CDS (Coding Sequence)
- **Protein Sequences**: Translated using NCBI Codon Tables
- **Quality Control**:
  - DNA sequences divisible by three in length
  - Start with a start codon
  - End with a single stop codon

## Dataset Structure

Each entry contains:
- DNA sequence
- Corresponding protein sequence
- Gene and organism information

## Uses and Applications

This dataset is valuable for various research areas and applications, including:
- Comparative genomics
- Codon usage analysis
- Protein expression optimization
- Synthetic biology and genetic engineering
- Machine learning models in bioinformatics

It has been used to train the CodonTransformer model for codon optimization tasks.

## Authors
Adibvafa Fallahpour<sup>1,2</sup>\*, Vincent Gureghian<sup>3</sup>\*, Guillaume J. Filion<sup>2</sup>‡,  Ariel B. Lindner<sup>3</sup>‡,  Amir Pandi<sup>3</sup>‡

<sup>1</sup> Vector Institute for Artificial Intelligence, Toronto ON, Canada  
<sup>2</sup> University of Toronto Scarborough; Department of Biological Science; Scarborough ON, Canada  
<sup>3</sup> Université Paris Cité, INSERM U1284, Center for Research and Interdisciplinarity, F-75006 Paris, France  
\* These authors contributed equally to this work.  
‡ To whom correspondence should be addressed: <br>
guillaume.filion@utoronto.ca, ariel.lindner@inserm.fr, amir.pandi@cri-paris.org
<br>

## Additional Resources
- **Project Website** <br>
  https://adibvafa.github.io/CodonTransformer/

- **GitHub Repository** <br>
  https://github.com/Adibvafa/CodonTransformer

- **Google Colab Demo** <br>
  https://adibvafa.github.io/CodonTransformer/GoogleColab

- **PyPI Package** <br>
  https://pypi.org/project/CodonTransformer/

- **Paper** <br>
  https://www.nature.com/articles/s41467-025-58588-7


## Citation
```
@article{Fallahpour_Gureghian_Filion_Lindner_Pandi_2025,
  title={CodonTransformer: a multispecies codon optimizer using context-aware neural networks},
  volume={16},
  ISSN={2041-1723},
  url={https://www.nature.com/articles/s41467-025-58588-7},
  DOI={10.1038/s41467-025-58588-7},
  number={1},
  journal={Nature Communications},
  author={Fallahpour, Adibvafa and Gureghian, Vincent and Filion, Guillaume J. and Lindner, Ariel B. and Pandi, Amir},
  year={2025},
  month=apr,
  pages={3205},
  language={en}
}
```