Temporal variability is a personalized feature of the human microbiome

Table 2 Measured factors that influenced the temporal variability of the human microbiome

	Parameter estimate	Sum of squares	F statistic	P value	BIC	R ²
Forehead - unweighted
Antibiotic use	0.015	0.010	8.76	0.004	-262.21	0.175
University	0.119	0.006	5.41	0.023	-263.38
Forehead - weighted
Median Shannon	0.038	0.090	32.2	3.61 e -7	-190.67	0.580
Gender	0.023	0.027	9.54	0.003	-193.54
Roommates (n)	0.039	0.016	5.70	0.02	-196.16
Gut - unweighted
Median Shannon	0.063	0.081	73.24	4.3 e -12	-240.90	0.570
Over-the-counter acne product	0.014	0.013	11.18	0.001	-249.97
University	0.014	0.007	6.64	0.012	-254.34
Gut - weighted
Median Shannon	0.107	0.238	20.64	2.61 e -5	-85.83	0.319
Over-the-counter acne product	0.034	0.065	5.65	0.021	-90.53
University	0.028	0.047	4.08	0.047	-90.55
Palm - unweighted
Exercise frequency	0.033	0.022	15.74	2.00 e -4	-188.6	0.310
Lives with dogs	0.014	0.010	7.18	0.009	-189.9
Roommates (n)	0.016	0.008	5.77	0.019	-191.8
Palm - weighted
Antibiotic use	0.026	0.024	4.97	0.029	-129.8	0.080
Tongue - unweighted
Antibiotic use	0.018	0.015	7.75	0.007	-217.82	0.215
Median Shannon	0.038	0.010	5.5	0.022	-220.12
Tongue - weighted
No good model

Generalized linear models (GLMs) were used to determine which of the measured factors or combination of factors best predicted variability in microbiome membership (unweighted UniFrac) and structure (weighted UniFrac). Unweighted UniFrac distances are a metric of the phylogenetic dissimilarity of samples through time. Weighted UniFrac distances weight dissimilarity both as a function of the phylogenetic dissimilarity and the relative abundance of taxa (such that two samples with the same phylogenetic dissimilarity are considered more different if one is dominated by a particular taxon).
BIC = Bayesian Information Criterion.

ISSN: 1474-760X