Discovering Alzheimer Genetic Biomarkers Using Bayesian Networks

Area of Science:

• Genetics and Genomics
• Computational Biology
• Neurodegenerative Diseases

Background:

• Single nucleotide polymorphisms (SNPs) are a major source of human genetic variation and are associated with complex diseases like Alzheimer's disease (AD).
• Identifying SNP biomarkers is crucial for the early diagnosis and treatment of AD.
• Bayesian networks (BNs) offer a framework for modeling gene and SNP interactions.

Purpose of the Study:

• To apply Bayesian network structure learning algorithms to whole genome sequencing (WGS) data.
• To detect causal SNPs and gene-SNP interactions associated with Alzheimer's disease.
• To identify novel SNP biomarkers for AD.

Main Methods:

• Whole genome sequencing (WGS) data analysis.
• Application of Bayesian network structure learning algorithms, including Markov blanket-based methods, Naïve Bayes, and Tree Augmented Naïve Bayes.
• Focus on polymorphisms within the top ten genes linked to AD by genome-wide association (GWA) studies.

Main Results:

• Several novel SNP biomarkers (rs7530069, rs113464261, rs114506298, rs73504429, rs7929589, rs76306710, and rs668134) were identified and found to be significantly associated with Alzheimer's disease.
• Bayesian network methods demonstrated effectiveness in identifying AD-causal SNPs with acceptable accuracy.
• Markov blanket-based methods outperformed Naïve Bayes and Tree Augmented Naïve Bayes, achieving higher accuracy (66.13%) and sensitivity (88.87%) compared to Naïve Bayes (61.58% and 59.43%).

Conclusions:

• Bayesian networks are effective tools for identifying causal SNPs and gene-SNP interactions relevant to Alzheimer's disease.
• The identified SNP set, particularly those found using Markov blanket methods, shows a strong association with AD.
• This study highlights the potential of BN in advancing the discovery of genetic biomarkers for improved AD diagnosis and treatment.