Abstract
The aging population worldwide is on the rise, leading to a higher number of Parkinson's disease (PD) cases each year. PD is presently the second most prevalent neurodegenerative disease, affecting an estimated 7-10 million individuals globally. This research aimed to identify mobile genetic elements in human fecal samples using a shotgun metagenomics approach. We identified over 44,000 plasmid contigs and compared plasmid populations between PD patients (n = 68) and controls (n = 68). Significant associations emerged between groups (control vs PD) based on plasmid alpha and beta diversity. Moreover, the gene populations present on plasmids displayed marked differences in alpha and beta diversity between PD patients and controls. We identified a considerable number of phage contigs that were differentially abundant in the two groups. We also developed a predictive machine learning model based on phage abundance data, achieving a mean Area Under the Curve (AUC) of 0.74 with a standard deviation of 0.105 and a mean F1 score of 0.68 with a standard deviation of 0.14 across cross-validation folds, indicating moderate discriminatory power. Additionally, when tested on external data, the model yielded an AUC of 0.74 and an F1 score of 0.8, further demonstrating the predictive potential of phage populations in Parkinson's disease. Further, we improved the continuity and identification of the protein coding regions of the phage contigs by implementing alternative genetic codes.