Deep Learning Framework Unveiled for Identifying Risk Genes and Drug Targets in Alzheimer’s Disease

A Cleveland Clinic-led research team has created an artificial intelligence (AI)-based framework for analyzing vast amounts of genetic data related to Alzheimer’s disease, opening up opportunities for target identification and drug discovery.

The tool, which is publicly available online at alzgps.lerner.ccf.org, is designed to push forward therapeutic discovery for Alzheimer’s disease, the most common form of dementia, which is expected to affect more than 150 million people by 2050. A study published by the research team in Cell Reports (2022;41[9]:11717) prioritizes 156 risk-associated genes through the framework, which is called NETTAG and is based in deep learning technology.

Developing AI technology-based tools is essential for taking advantage of genomic sequencing data as Alzheimer’s disease rates continue to climb, says the study’s corresponding author, Feixiong Cheng, PhD, assistant staff in Cleveland Clinic’s Genomic Medicine Institute.

“Advances in capable and intelligent computer-based algorithms offer the opportunity to harness large-scale data to pinpoint functional variants and risk genes that drive Alzheimer’s disease,” he says. “That allows us to identify targets for treatment development.”

Genetic sequencing data is becoming more readily available through projects like the Alzheimer’s Disease Sequencing Project funded by the National Institute on Aging. National Institutes of Health-funded AI and machine learning programs share genetic and genomic information to aid research efforts for accelerating Alzheimer’s treatment development, Dr. Cheng notes.

NETTAG essentials

This study integrated information from large-scale Alzheimer’s disease genetic databases and a Cleveland Clinic-developed interactome, which describes fundamental interactions between proteins in human cells. The researchers used NETTAG, which stands for network topology-based deep learning framework to identify disease-associated genes, to analyze the data in search of potential drug targets and repurposable treatments for Alzheimer’s disease.

The genetic sequencing information from public databases provides data on which genes are likely to be associated with Alzheimer’s disease. Connecting this information with the interactome then identifies the potentially targetable biological pathways associated with these genetic markers.

Gemfibrozil flagged as a candidate drug

The researchers chose four drugs for further screening based on the patient analysis: ibuprofen, gemfibrozil, cholecalciferol and ceftriaxone. The study outlines multiple methods for gauging the potential of each drug: verifying outcomes against millions of patients’ electronic health records, comparing similar drugs in clinical trials, and analyzing race- or sex-specific outcomes.

Those analyses identified gemfibrozil, a cholesterol medication prescribed to reduce elevated serum triglyceride levels, as a strong candidate drug for potential prevention and treatment of Alzheimer’s disease.

“Patient records showed that gemfibrozil is associated with significantly lower Alzheimer’s disease risk, and we have the genetic and protein-protein data to show it targets molecular pathways associated with the disease,” Dr. Cheng explains. “This method provides strong evidence to rapidly identify candidate drugs for clinical trials, but it can also serve as validation for drugs identified through other methods, accelerating the drug discovery process.”

Broader AI research underway

Funded by the National Institute on Aging, Dr. Cheng’s team is developing and using AI and deep learning technologies to support creation of innovative tools capable of analyzing 10,000+ sequenced whole-genomes available from the Alzheimer’s Disease Sequencing Project. They will then use the AI tools to identify novel drug targets and molecular networks involved in AD as well as genetic evidence-supported repurposable medicines.