Cindy Boer

The Gut Microbiome of Childen and Adults | 209 5.1 Technical covariates in the initial stool 16S datasets of the two cohorts After generating the initial 16S datasets of both the RS and GenR cohorts, the effects of collection time (time in mail; TIM) and the yield of DNA isolation runs (Batch; Supple- mentary Figure 3 ) on the 16S profiles were assessed. To identify the impact of these technical covariates on the α-diversity and overall profiles, Shannon α-diversity and Bray-Curtis dissimilarity metrics were calculated in each cohort using ‘diversity’ and ‘vegdist’ functions in vegan, respectively. The effects of the covariates on α-diversity were assessed by linear regression ‘lm’ function in package stats. The basic model was adjusted for age and sex. Each of the two technical covariates were stepwise added in order to identify its contribution to the model. Then, likelihood ratio test was used to compare these nested models. Additionally, in GenR self-reported ethnicity was added to the model to inspect and compare the effect of this covariate. Subsequently, permu- tation analysis of variance (PERMANOVA) was performed to inspect the global effects of these technical covariates on the overall profiles using the ‘adonis’ function in vegan. Furthermore, for TIM, additive general linear regression analyses of single genus-level OTUs were performed in MaAsLin. For this, we used sub-sampled datasets that included only samples up to a certain TIM (up to 7 days). Regression analyses in MaAsLin were performed after arcsine square root transformation of the relative abundance data and were adjusted for age, sex and BMI to assess the effect of TIM exclusively. Validation of the final stool 16S datasets of GenR and RS Final datasets were constructed by excluding samples that had been in the mail for 3 (in RS) or 5 (for GenR) days (see Results section). Average relative abundances of the six major phyla were compared to those of other large cohorts. GenR profiles were compared to those reported by the Copenhagen Prospective Study on Asthma in Child- hood (COPSAC) cohort (Copenhagen, Denmark; n=156; age range=4–6 years[39]), and RS profiles were compared to those reported by the Dutch LifeLines-DEEP (LLD) cohort (Groningen, the Netherlands; n=1,135; age range=20–90 years[40,41]) and the Belgian Flemish Gut Flora Project (FGFP) cohort (Leuven, Belgium; n=1,106; age range=20–80 years[42]). In addition, we performed association analyses of BMI with Shannon α-diversity and single OTUs at genus level to confirm this well-established as- sociation11,13,[40] in our cohorts, and by comparing associated OTUs observed in our cohorts to those reported by the above-mentioned cohorts. The LifeLines DEEP (LLD) studywas approved by the ethics committee of the Uni- versity Medical Centre Groningen, document number METC UMCG LLD: M12.113965 and all methods were performed in accordance with the Declaration of Helsinki. All participants signed an informed consent form prior to study enrolment.