Cindy Boer

212 | Chapter 5.1 7. In contrast, we did not observe any substantial changes in phylum-level profiles with respect to the DNA-Batch variable ( Supplementary Figure 4 ). Moreover, TIM had a small but significant negative effect on α-diversity in GenR and RS (beta=−0.02 alpha units/day, p-value=9.3x10 -03 and beta=−0.03, p -value=4.6x10 -03 , respectively). Again, we did not observe an effect of Batch on α-diversity. Correlations of TIM and Batch with overall composition (β-diversity) in both cohorts were small but significant (R 2  =0.004, p-value=0.001 and R 2  =0.002, p-value=0.005 for TIM and R 2  =0.01, p-value=0.001 and R2=0.005, p-value=0.001 for Batch in GenR and RS, respectively). In taxonomy-based analyses (MaAsLin) at genus-level of both datasets for TIM, we observed increased abundances of genus Escherichia/Shigella upon prolonged times in the mail ( Figure 1 ). However, these differences were only significant after 3 days in the RS cohort and after 5 days in the GenR cohort ( Figure 1 ). Furthermore, we observed smaller decreases in the abundances of a number of genera, including Roseburia and Coprococcus , upon prolonged TIM. In order to further evaluate the importance of TIM and Batch effects, we added them to linear models for the analysis of well-established association of α-diversity with BMI[11,13,40] in both GenR and RS. Although, the estimates (betas) o f the exposure (i.e., BMI) remained similar after various levels of adjustment, the model including sex, age and TIM (model 3) as co-variates provided a better fit to the data than the model including only sex and age (model 2; likelihood ratio test: p-value=5.8x10 -03 and p -val- ue=5.5x10 -03 for GenR and RS, respectively). Including Batch, within model 1 (model 2), resulted in a better fit to the data in RS only ( Table 1 ; p-value=1.9x10 -04 ). Table 2: association of BMI with Shannon diversity in the 16S datasets of GenR and RS cohorts. Model Linear model: α-diversity ~BMI + covar. R 2 Estimate P-value R 2 Estimate P-value 0 BMI 0.008 -0.031 9.6x10 -06 0.015 -0.021 9.9x10 -07 1 BMI + sex 0.008 -0.031 7.3x10 -06 0.014 -0.021 1.0x10 -06 2 BMI + sex + age 0.008 -0.032 7.0x10 -06 0.015 -0.021 7.8x10 -07 3 BMI + sex + age + TIM 0.011 -0.032 1.0x10 -05 0.019 -0.021 9.2x10 -07 4 BMI + sex + age + TIM + Batch 0.011 -0.032 1.0x10 -05 0.027 -0.020 1.6x10 -06 5 BMI + sex + age + TIM + Batch + ethnicity 0.38 -0.021 3.2x10 -03 Stepwise linear model used for each covariates in the analysis of association of microbial diversity with BMI (TIM: time in mail).