Understanding the complex metabolic changes that occur in Alzheimer’s disease (AD) is essential for developing better diagnostic and therapeutic strategies. Traditionally, researchers use genome-scale metabolic models (GEMs) to map gene expression data to metabolic pathways, helping identify disease-related disruptions. However, these models often overlook another crucial factor—genetic variants that alter enzyme functionality.

Now, a research team led by researchers at Gebze Technical University has introduced an innovative approach that integrates both gene expression data and genomic variants from RNA sequencing to create more accurate, personalized metabolic models for Alzheimer’s disease.

What Are Genome-Scale Metabolic Models

GEMs are computational models that represent all known metabolic pathways in an organism. Scientists use these models to predict how cells process nutrients, generate energy, and maintain homeostasis. When gene expression data from RNA sequencing (RNA-seq) is mapped onto GEMs, it provides a snapshot of metabolic activity in healthy and diseased states.However, metabolic function is not solely determined by gene expression—genetic variants can also disrupt enzyme activity, affecting metabolic pathways. By incorporating pathogenic genetic variants identified in RNA-seq data, this new approach improves the accuracy of GEMs, helping uncover hidden metabolic disruptions in AD patients.

How Does This New Approach Work?

The researchers extracted both transcriptomic (gene expression) and genomic (mutation) data from the same RNA-seq dataset. They then mapped genes with a high load of pathogenic variants found in AD patients onto a human GEM. The personalized patient metabolic models were compared to control models, revealing key findings:

1. Improved Accuracy – The new method more precisely detected AD-associated metabolic pathways than models based solely on gene expression.
2. Uncovered Hidden Pathways – Several metabolic pathways that were previously overlooked in AD were identified when genetic variants were taken into account.
3. Better Insights into Disease Mechanisms – The study highlights how combining gene expression and genetic variants can provide a more comprehensive view of metabolic dysfunction in Alzheimer’s disease.

Summary of iMAT-based personalized metabolic model generation and incorporation
of the impact of pathogenic variants to models

A RNAseq data are obtained in Fastq format and processed with two different pipelines to find gene expression values and genomic variants. The genes with higher load of pathogenic variants and the genes with high or moderate expression levels were selected as inactive genes for each AD sample. B Gene expression values are mapped to the genome-scale metabolic model and AD, control and AD^var models are created with the iMAT algorithm. To create AD^var models, unlike the classical method, the genes carrying higher load of pathogenicity in AD were additionally used. C Differentially perturbed reactions between AD and control are identified by Fisher’s Exact Test after representing each personalized metabolic model as a binary vector, where “1” means the reaction is active in that model. Then, pathways overrepresented with the differentially perturbed reactions were predicted.

Implications for Alzheimer’s Research and Treatment

This integrated approach could revolutionize personalized medicine by creating tailored metabolic models for individuals, leading to:

• More precise biomarkers for early diagnosis
• Better understanding of metabolic dysfunction in AD
• New therapeutic targets based on patient-specific metabolic alterations

By harnessing the full potential of RNA sequencing, this research opens new doors for studying Alzheimer’s disease at both the transcriptomic and genomic levels, paving the way for more effective treatments.