ASO Atlas 2.0: 283,239 Antisense Oligonucleotide Measurements

ASO Atlas 2.0 is a benchmark dataset containing 283,239 measurements across 168,395 unique antisense oligonucleotide compounds. The data was extracted from 606 USPTO patent filings using an LLM-based pipeline and is encoded in HELM notation for direct model input. It was created by barneyhill and last updated on April 30, 2026.

Use Cases

• Benchmarking ASO prediction models based on the four preclinical assay types described.
• Training machine learning models for compound property prediction based on HELM-encoded chemical structures.
• Analyzing structure-activity relationships for antisense oligonucleotides based on the patent-derived compound library.

Strengths

• Contains 283,239 measurements across 168,395 unique compounds.
• Compounds are encoded in HELM notation, enabling direct use as model input.
• Data is extracted from 606 USPTO patent filings, suggesting a real-world industrial context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the last update is dated in the future (2026-04-30).

Provenance

Source

Extracted from 606 USPTO patent filings.

Collection Method

LLM-based extraction pipeline.

Freshness

Last updated 2026-04-30 20:55:05