GOLD standard dataset for Alzheimer genes

Area of Science:

• Genetics
• Neurodegenerative Disorders
• Bioinformatics

Background:

• Alzheimer disease (AD) is a complex, multigenic disorder involving numerous genes.
• Existing genetic association databases contain numerous false positives and negatives, limiting their utility.
• Hundreds of susceptible genes are implicated in AD development and progression.

Purpose of the Study:

• To compile a comprehensive and validated list of Alzheimer disease-associated genes.
• To classify gene associations as positive, negative, or ambiguous.
• To create a gold standard dataset for machine learning model training.

Main Methods:

• Utilized the Genetic Association Database as a reference.
• Performed double-fold cross-validation on the existing data.
• Extracted gene associations, classification (positive, negative, ambiguous), and supporting sentences.

Main Results:

• Generated a curated list of Alzheimer disease genes with validated association classes.
• Established a comprehensive reference dataset for Alzheimer's disease genetics.
• The dataset includes reference sentences confirming gene-disease associations.

Conclusions:

• The generated dataset serves as a gold standard for training machine learning classifiers.
• This resource will improve prediction accuracy for gene associations in Alzheimer and other diseases.
• Positive-associated gene data can support system-level modeling and meta-analyses in Alzheimer research.