AMELIE

AMELIE automates literature evaluation to identify publications linking genetic variants and phenotypes for the diagnosis of Mendelian disorders.

Key Features:

• Comprehensive Literature Parsing: Processes all 29 million PubMed abstracts and hundreds of thousands of full-text articles to extract evidence of variant causality and phenotype associations.
• Prioritization Algorithm: Ranks patient candidate variants and genes by their likelihood of explaining a given set of patient phenotypes.
• Efficiency in Diagnosis: In a study of 215 diagnosed singleton Mendelian patients AMELIE ranked the causative gene first in two-thirds of cases and achieved rapid diagnosis for over 90% of cases by evaluating only the top 11 genes out of a median of 127 candidate genes per patient.
• Comparative Efficiency: Demonstrates a 3- to 19-fold increase in evaluation efficiency versus traditional hand-curated database approaches, with replication in clinical case cohorts from Stanford Children's Health and the Manton Center for Orphan Disease Research.

Scientific Applications:

• Clinical diagnosis of Mendelian disorders: Prioritizes candidate genes and supporting publications to accelerate variant interpretation for clinicians, including singleton patients without relatives' exomes.
• Genotype–phenotype evidence extraction: Aggregates literature evidence linking specific genetic variants to phenotypes from PubMed abstracts and full-text articles.
• Comparative evaluation of diagnostic workflows: Provides ranked gene lists used to compare literature-based prioritization efficiency against hand-curated database approaches in clinical cohorts.

Methodology:

Automated parsing of PubMed abstracts and full-text articles to extract mentions of genetic variants and associated phenotypes, followed by a prioritization algorithm that ranks candidate genes/variants by their likelihood of explaining patient phenotypes.