SciGroveBeta
Medicine

Affinage: genome-scale mechanistic gene annotation from the published literature

Matteo Di Bernardo, Iain M. Cheeseman

Featured July 30, 2026

Original Paper

This analysis was generated by SciGrove. Upload your own PDFs or enter a DOI — and get the same AI breakdown on any paper.

Get started

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

A new computer program uses smart AI to read all the science papers about genes, picking out only real experiments to explain what each gene actually does, making a super-updated and reliable gene dictionary for everyone.

In depth
The paper introduces Affinage, a novel two-stage Large Language Model (LLM) pipeline designed to perform genome-scale mechanistic gene annotation directly from primary literature. Unlike existing curated databases that lag behind or naive LLM approaches that overclaim, Affinage uses biologist-designed prompts to strictly extract only direct experimental evidence in a reproducible manner. This evidence is then synthesized into structured, literature-grounded records for nearly all human protein-coding genes, significantly improving coverage and recency compared to traditional methods.

Key Takeaways

  • 1
    Affinage employs a two-stage LLM pipeline (reading and synthesis) to generate structured, mechanistic gene annotations directly from primary literature, bypassing the limitations of manual curation and naive LLM use.
  • 2
    The pipeline incorporates biologist-designed prompts for strict evidence extraction, ensuring that only direct experimental findings are used, preventing overclaiming and enhancing the reproducibility and faithfulness of annotations.
  • 3
    The resulting genome-scale resource covers 19,293 human protein-coding genes, providing mechanistic descriptions for thousands of genes previously uncharacterized or poorly described in curated databases like UniProt, with a 99.1% win rate over decided pairs in quality evaluations.

Conceptual Flow

HIGH LEVEL
1
Methodology: Building a Reliable Gene Dictionary

The program first finds all relevant papers, then uses a smart AI to read them carefully for facts, and finally another AI summarizes those facts into a clear explanation for each gene.

All Science Papers
Find Relevant Papers
Gene-Specific Papers
2
Results: More Genes Explained Better

The new program explains many more genes than old methods, and its explanations are more detailed and up-to-date, helping scientists understand how our bodies work.

Old Gene Explanations
Add New Details
Richer Gene Explanations