Alzheimer's disease (AD), the most common cause of dementia and impaired cognitive function, still has no effective treatment, according to researchers. So research is centered on identifying AD biomarkers and targets.
Now, scientists at King Abdullah University of Science and Technology in Saudi Arabia have created a computational method that identifies AD biomarkers and targets. It combines multiple "hub gene" ranking methods and "feature selection" methods with machine learning and deep learning to identify hub genes and gene subsets, the researchers used three AD gene expression datasets using six ranking algorithms and two feature-selection methods.
The researchers then created machine learning and deep learning models to identify the gene subset that best distinguished Alzheimer's disease samples from healthy controls. They found that feature selection methods outperformed hub gene sets in terms of prediction performance and that the five genes identified by both feature selection methods had an "AUC" of 0.979.
Based on a literature review, the researchers further showed that 70% of the upregulated hub genes (among the 28 overlapping hub genes) were AD targets, with six miRNA and one transcription factor associated with the upregulated hub genes. According to the researchers, overlapping upregulated hub genes can narrow the search space for potential novel targets.
Source: Nature Scientific Reports (open-access)
Images: MidJourney, Prompts by Lewis Farrell
Interesting story? Please click on the ? button below!