A comprehensive dataset of 188,521 RNase H-mediated antisense oligonucleotides (ASOs) with experimentally validated efficacy measurements, extracted from 417 USPTO patents published between 2001 and 2025.
ASO Atlas provides the first large-scale, publicly available resource for understanding the relationship between ASO sequence, chemical modifications, and inhibitory activity. The dataset includes:
- 188,521 gapmer ASOs targeting 306 human genes
- Sequence composition and target genomic locations
- Chemical modification patterns (phosphorothioate backbones, 2'-MOE, cEt sugar modifications)
- Quantitative efficacy measurements from qRT-PCR experiments across multiple cell lines
- Experimental metadata including cell type, dosage, and transfection method
# Clone the repository
git clone https://github.com/barneyhill/aso_atlas.git
cd aso_atlas
# Create and activate conda environment with Python 3.11
conda create -n oligoai python=3.11
conda activate oligoai
# Install dependencies
pip install -r requirements.txtimport pandas as pd
# Load the dataset
df = pd.read_pickle('data/aso_atlas.pkl')
# View schema
print(df.columns)
# ['aso_sequence_5_to_3', 'inhibition_percent', 'chemistry', 'custom_id',
# 'target_mrna', 'target_gene', 'cell_line', 'dosage',
# 'cells_per_well', 'transfection_method', ...]For a complete walkthrough, see the demo notebook.
- Preprint: https://www.biorxiv.org/content/10.1101/2025.10.29.685292v1
- OligoAI Model: https://huggingface.co/barneyhill/OligoAI
- OligoAI Training Code: https://github.com/barneyhill/OligoAI
- Web Interface: https://sitlabs.org/OligoAI