Brain virome dysbiosis in Parkinson’s disease and multiple system atrophy

Figures

• 

  Figure 1. Brain virome composition profiles and core brain virome. (A) Pie chart depicting the viral species distribution at the phylum level in HCs and NDs. (B) Pie chart illustrating the family-level distribution of viral species in HCs and NDs. (C) Heatmap of the core virome of HCs and NDs at the genus level. (D) Heatmap of the core virome of HCs and NDs at the species level.

• 

  Figure 2. The brain virome richness and diversity and interpersonal variations in HCs and NDs. (A) Boxplot of alpha diversity of Observed and Chao1, Shannon and Simpson’s indices reflect the abundance and diversity of OTU in the samples. (B) PERMANOVA-validated non-metric multidimensional scaling (NMDS) beta diversity depicted with Bray-Curtis and Jaccard index distances.

• 

  Figure 3. Brain viral signatures are significantly different between HCs and NDs (A) Cladogram of the LEfSe analysis of the brain virome, with taxa enriched in HCs in green and those enriched in NDs in red. (B) Linear discriminant analysis (LDA) effect size analysis (LEfSe) identified the most differentially abundant viral species between HCs and NDs (p < 0.05; LDA score 2). HCs-associated viral species are indicated with negative LDA scores (green) while NDs-associated viral species are indicated with positive LDA scores (red color). (C) Mann–Whitney test for each of the identified viral signature displayed (p < 0.05). (D) Receiver operating characteristics (ROC) analysis showed that the area under curves (AUC) values of the individual viral signature differ between HCs and NDs (AUC; 95% CI).

• 

  Figure 4. Evaluation of the predictive power of viral signatures identified between healthy controls (HCs) and neurodegenerative disorders (NDs). (A) The heatmap represents the co-occurrence of viral signatures identified by Spearman correlation analysis. (B) The CombiROC curve combines multiple biomarkers and displays their discriminatory power between the two compared classes. (C) The pie charts illustrate the fraction of predictions, including false negatives (FN), false positives (FP), true negatives (TN), and true positives (TP), providing an overall assessment of the predictive performance of the model. (D) The violin plot depicts the data’s probability density for HCs and NDs based on the previously determined optimal cut-off on the corresponding ROC curve.

• 

  Figure 5. Validation of brain virome signatures in independent cohorts. (A) Heatmap showing prevalence of selected viral taxa in the PD validation cohort (PRJNA555099). (B) Venn diagram depicting overlap between the discovery cohort (32 striatal samples) and validation cohort (10 PD samples), with 16 viral taxa consistently shared. Bar plots below show the number of unique and overlapping species. (C) List of common viral taxa detected in both discovery and validation cohorts, including HERV-K, murine osteosarcoma virus, woolly monkey sarcoma virus, and Saccharomyces cerevisiae killer virus M1. (D) Viruses identified by LEfSe in the discovery analysis and their detection status in the validation cohort. Together, these analyses demonstrate that a reproducible set of 16 viral taxa is consistently detectable across independent striatal datasets, supporting the existence of a putative core brain virome rather than random or project-specific artifacts.