Table 1.
Closely age-matched alpha- and beta-diversity results.
Table 2.
Taxa proposed as elevated/reduced in ADHD.
Table 3.
Cohort analysis.
Fig 1.
Alpha-diversity of Control (blue) and ADHD (red) samples using (in order) the count of unique taxa, Chao1 richness [102], and phylogenetic diversity, with error bars.
Fig 2.
Beta-diversity of Control (blue) and ADHD (red) samples computed using unweighted and weighted Unifrac [104] distance.
Fig 3.
Results of running sPLS-DA [76] on microbiome abundance data (ellipse confidence level 95%), comparing Control (orange) and ADHD (blue) groups.
Table 4.
Principal component error rates.
Fig 4.
Differential abundance (Cladogram).
Taxa reported by LEfSe as elevated for Control (orange) and ADHD (purple) groups, produced by LEfSe [77] (p ≤ 0.05, Bonferroni correction). LEfSe-reported taxa are plotted on a cladogram, with each concentric circle representing a phylogenetic classification level (innermost = phylum). Shared areas represent distinctive regions of the phylogenetic tree.
Fig 5.
Differential abundance (Bar graph).
LEfSe-reported taxa ordered by Linear Discriminant Analysis (LDA, [122]). A higher magnitude indicates more reliable differentiation.
Table 5.
Taxa reported by various differential abundance methods throughout our analysis.
Fig 6.
Plots of taxa relative abundance for each subject, generated using Qiime [81], conducted at the phylum level. Subjects are ordered by increasing Adult ADHD Self Report Scale (ASRS) score, with the y-axis representing relative abundance.
Fig 7.
Plots of taxa relative abundance for each subject, generated using Qiime [81], conducted at the genus level. Subjects are ordered by increasing Adult ADHD Self Report Scale (ASRS) score, with the y-axis representing relative abundance.
Fig 8.
Upper-level Microbial Co-occurrence Networks (MCNs).
MCNs at the phylum (A), class (B), and order (C) taxonomic levels, visualized using Cytoscape [57], and oriented by Fruchterman-Reingold [58]. Nodes represent taxa, colored by phylum (yellow = Firmicutes, purple = Bacteroidetes, brown = Actinobacteria, blue = Proteobacteria) with size directionally proportional to abundance. The co-occurrences are distinguished by those that co-habit (green edges) and co-avoid (red edges). SparCC [112] correlation (p≤0.01) was used as edge weight and also the parameter for Fruchterman-Reingold when determining edge length (larger = closer). SparCC correlations are shown at the phylum level. All taxa found as important by ATria are denoted by a pound sign (#) followed by its rank (ties indicated).
Table 6.
Upper-level taxa correlations.
Table 7.
Upper-level ATria rankings.
Fig 9.
MCNs at the family (A), genus (B), and species (C) taxonomic levels. Network visual properties, including node and edge size, color, and orientation, are the same as Fig 8. Taxa noted throughout our analyses are labeled (* = family-level taxon, # = genus-level taxon).
Fig 10.
Same MCNs as Fig 9, after clustering with the affinity propagation (AP) algorithm [85]. Family-level clusters are each given a unique color, and labeled with their dominant phylum and member family. New clusters that form at each lower taxonomic level are labeled, colored with shades corresponding to their dominant phylum/family when applicable—i.e. at the genus level FL1-FL3 are different shades of gold (family-level FL). Taxa noted throughout our analyses are labeled (* = family-level taxon, # = genus-level taxon).
Fig 11.
Heatmap representation of taxa correlations (green = positive, red = negative), with taxa organized on each axis by cluster (symmetric matrix). The area corresponding to the intersection of each cluster with itself is outlined with a box using the corresponding cluster color in Fig 10. Taxa and clusters noted throughout our analyses are labeled on the axes.
Table 8.
Burkholderiales-Bacteroidales correlations.
Table 9.
Cluster BB size.
Table 10.
Bacteroidaceae/Bacteroides dominant clusters.
Table 11.
Bacteroidaceae-dominant community intra- and inter-correlations.
Table 12.
Cluster BB centroid connectivity.
Table 13.
Bacteroidetes rankings.
Table 14.
Negative correlations between cluster BB and other clusters.
Table 15.
Additional Bacteroidetes observations.
Table 16.
Bifidobacterium correlations.
Table 17.
Bifidobacterium rankings.
Table 18.
Proteobacteria-Firmicutes correlations.
Table 19.
Mixed-family clusters.
Table 20.
Mixed-family cluster member rankings.
Table 21.
FT members.
Table 22.
Coriobacteriaceae rankings.
Table 23.
Adlercruetzia correlations.
Table 24.
Bacteroidetes-Firmicutes positive correlations.
Table 25.
Streptococcaceae/Streptococcus rankings.
Table 26.
Blautia and Oscillospira rankings.
Table 27.
Blautia and Oscillospira clusters.
Table 28.
Lachnospiraceae and Ruminococcaceae rankings.
Table 29.
Largest, consistently present clusters.
Table 30.
Cluster member interactions.
Table 31.
Taxa noted throughout our analysis.
Table 32.
Relationships noted throughout our analyses.