One SNP on chromosome 7 in the gene IMMPL2 reached genome-wide significance. Another gene IINHBA-AS1 on chromosome 7 achieved genome-wide significance when analyzed by KGG4 that relies on a composite association score including all SNPs in each known gene. The significance of these associations was not influenced by “p-hacking” statistical biases common in GWAS because phenotype choice was not based on previous association tests. This approach is a model for using heritability to reduce the multiple testing problems seen in many GWAS reports and it could be the method of choice in the design of GWAS studies in which sample size may be limited. Bray-Curtis, Weighted UniFrac, and to a lesser extent Unweighted UniFrac β-diversity demonstrate that many components of the microbiome community are heritable . While a shared environment and behavioral habits contribute to a more similar microbiome , such studies did not control well for the clear genetic influences in their populations. When we examined the differences among MZ and DZ cotwins and age-matched unrelated individuals that we were confident cohabitated , the genetic influences remain clear. It is significant that the genetic effects are detected using measures that include all detectable OTUs. To assess heritable influences of individual microbial components, we carried out intraclass correlation analyses that show that heritability extends across nearly all observed taxa individually . The one exception is in the fusobacteria where ICC does not distinguish MZ and DZ.
Possibly these organisms, known to be “bridges” between early and late colonizers on gum and tooth surfaces,drying rack cannabis may not have interaction with host proteins and could lack human genetic influences. GWAS of complex traits on relatively small samples is problematic due to the lack of statistical power. The influence of individual genes on traits that have multiple genetic components may be small. Moreover, the microbiome is a highly complex population with interacting networks of bacteria that all may have multiple interactions with the host. A variety of covarying network modeling approaches have demonstrated how complex these communities are. It has been shown that assuming the number of causal variants and their frequency spectra for a pair of traits are similar, more heritable traits are more likely to be detectable in GWAS. Therefore we focused on those microbiome endophenotypes with greatest additive genetic heritability for GWAS. Both ACE/ADE modeling and GCTA SNP heritability are suited to this approach. The microbial phenotypes with greatest additive genetic influence in the ACE/ADE model on the entire twin cohort were the abundance of the OTU4483015 that corresponds to an unnamed species of Granulicatella and PCo2 of Bray Curtis . The influence of additive genetics was variable depending on the trait when comparing the full sample to heritability only among cotwins that cohabitate . The variation in estimates may reflect environmental effects or loss of power between the full sample and the cohabitating sample . This again points to the complex nature of the microbe-host interactions in primarily aerobic and anaerobic environments and how human genetic influences must also be complex. As a further test of heritability prior to GWAS, we examined SNP-based heritability in our unrelated sample with GCTA. A positive correlation was observed between the ACE/ADE and GCTA ‘heritability’ estimates for continuous traits in both the full twin sample and the EUR sample .
Previous studies have demonstrated that large samples are needed to produce results reaching statistical significance using GCTA. In their original paper Yang et al. showed that while increasing the sample size does decrease the error bars of the heritability estimates, the heritability estimates themselves remain relatively stable. While the GCTA estimate was not significant upon correction for multiple testing, the positive correlation between the unrelated individuals and the twin studies provides support for the conclusion that for these continuous traits genetic variation influences microbial populations. A GWAS analysis with the six most heritable continuous traits determined from the twin modeling was carried out in the European populations . The GWAS of the abundance of the genus Granulicatella identified a genome wide significant SNP on chr7 . This SNP is located in an intron of the IMMP2L gene. The GWAS meta-analyses combining the EUR and ADM samples using METAL with the same 6 traits showed no new information about the chr7 SNP due to its low frequency in the ADM population but did produce an additional association with suggestive significance, chr12:82,166,911 for the phenotype Unweighted UniFrac PCo3, though it was not robust to correction for multiple testing. This SNP is located in the gene LIN7A that is widely expressed in endothelial cells. Markers in LD with the top SNPs were also highly associated with the phenotype, but in addition, markers of somewhat lower LD that were nearby also displayed elevated significance for both hits. This provides an argument that these loci may not be due purely to chance . To be adequately powered one must have a large sample size or the single SNP effect must be very large. However, most complex traits are polygenetic and so many loci with small effects account for the variation of the trait. Therefore, where sample size is limited, it may be difficult to observe significant SNP associations. To address this, it is possible to use biological information to inform analyses and increase statistical power. This may be done by aggregating the association of multiple SNPs known to be present within a known gene.
By this approach, the possibly small effects of all SNPs in the gene are combined and then the association of the entire gene may be determined. Even if no single SNP is found to be genome-wide significant the combined SNP contributions across the gene may be. One widely used gene-based GWAS analysis method is the Knowledge-based mining system for Genome-wide Genetic Studies.An analysis by KGG4 confirmed the gene IMMP2L and additionally identified a non-coding RNA INHBA- AS1 as significant hits. IMMP2L functions in the mitochondrion where it is involved with processing of signal peptides as a peptidase directing transport to the interior mitochondrial space. INHBA- AS1 and INHBA were previously associated with dental caries in a GWAS, and INHBA was postulated to influence the development of dental caries via its role in tooth morphology development. In support of this hypothesis Zeng et al. discuss that INHBA has been shown to be important for tooth development and knockout mice of INHBA have alterations in the eruption of new teeth. Attachment to the tooth surface is a part of the establishment of the oral microbiome and disruption of this process could lead to changes in the community structure of oral biofilms. Ascribing functional significance to IMMP2L, INHBA-AS1, or LIN7A, is speculative in the absence of a replication experiment. Nevertheless, this study is among the first to use heritability to refine microbiome phenotypes prior to GWAS testing and the findings will provide a basis for additional genetic studies in larger replication samples and in future molecular analyses. Of the 100 most significantly associated SNPs for each of the 6 GWAS analyses in the EUR sample, 7 SNPs were shared at least twice among Bray Curtis PCo2, Unweighted UniFrac PCo2, and Weighted UniFrac PCo2 analyses probably due to shared underlying variation of PCo2. A comparison of SNPS from the Granulicatella GWAS and the PCo3 unweighted UniFrac MetaAnalysis in our experiments with other published GWAS studies of the microbiome found that the majority of overlapping SNPs followed a normal distribution, commercial greenhouse supplies and those few that did deviate from expectation did not reach genome wide significance in either study. It is perhaps not surprising that genes showing influence in gut do not appear in salivary samples. There is very little overlap in organism composition between niches and it can be argued that one reason for this is that different genes influence each niche. Genes and environment potentially contribute to all aspects of the microbiome. Whereas twin studies are particularly powerful in differentiating between them, GWAS is poorly suited to teasing these factors apart. We show that tobacco/marijuana/alcohol use has little influence on the ability to detect associations of our top scoring loci. This is somewhat unexpected in that it is well known that some microbes either increase or decrease in response to tobacco. This is consistent with a hypothesis that the tobacco effects seen are mostly free of significant genetic influences and that conversely, the genetic effects we find do not dependent on environmental perturbations to be observed.
The results point out a need for well-controlled gene by environment experiments to fully understand how genes work and how environmental factors actually influence microbial communities.Parents hold influential roles in preventing youth tobacco use. Children of nonsmoking parents are less likely to initiate tobacco smoking themselves,and interventions that actively engage parents have revealed promise in reducing youth tobacco, alcohol, and illicit substance use.8 Creating tobacco-free home environments is one approach parents may take to set norms and expectations about tobacco use.9 Children and/or young adults who live in households with strict rules against smoking are less likely to try or regularly smoke cigarettes.Setting household rules for all family members and visitors may be more effective than invoking tobacco use rules applicable only to children, which are not necessarily associated with less youth smoking.Talking explicitly to children about not using tobacco represents another possible approach to discourage tobacco use. Parent-child antitobacco communication has been associated cross-sectionally with greater quit intentions among youth already using tobacco but has been inconsistently associated with cigarette smoking among youth overall.For both household rules and talking about tobacco, existing studies are predominantly cross-sectional and focused only on cigarette smoking. Parents face new challenges in addressing youth substance use in a changing tobacco landscape. Although youth cigarette smoking is declining, use of non-cigarette products, notably electronic cigarettes , is sharply increasing.Often small, sweet smelling, and unfamiliar to parents in appearance, e-cigarettes may be easier than cigarettes for children to conceal, possibly contributing to less parental awareness about youth use.18 Even for cigarettes, previous findings suggest parental awareness lapses. In one study, among adolescents who smoked 1 to 5 cigarettes a day, only 39% of parents were aware of their use.In another, 43% of parents correctly identified that their child had smoked a cigarette within the last 6 months.In the current study, we consider parental knowledge or suspicion, household rules, and talking with youth about tobacco in a nationally representative longitudinal study of youth. Specifically, using data from the youth component of the Population Assessment of Tobacco and Health Study, we addressed 2 main research questions using 2 different analytical approaches. First, we conducted a time-series analysis using 4 waves of PATH Study youth data to assess parent or guardian knowledge or suspicion that their child uses tobacco. Next, we estimated longitudinal associations of household rules and talking to children about tobacco with youth tobacco initiation over 1, 2, and 3 years. We aim to characterize lapses in parental awareness and to evaluate potential parental strategies to prevent youth tobacco initiation.An are aprobability, 4-stage stratified sampling design was implemented at wave 1 to represent the US non-institutionalized civilian population. Parents were asked about their participating children separately from questionnaires administered to youth directly. The weighted wave 1 response rate for the youth survey was 78.4% among households screened for participation.Respondents were followed annually in waves . Respondents who reached age 18 before follow-up were invited to join the adult component. “Shadow youth,” aged 9 to 11 at wave 1, were invited to join the study at the wave at which they reached age 12. An additional “replenishment” youth sample was enrolled at wave 4.PATH Study investigators obtained a National Institutes of Health certificate of confidentiality and ethical approval from the Westat Institutional Review Board. Parents or guardians provided informed consent for assenting youth. Youth received $25 for participation and parents or guardians received $10. In the present analysis, we used fully deidentified public use files.Each wave, youth responded separately about lifetime use and past 30-day use of multiple tobacco products. For this analysis, tobacco use was categorized as never use, former use , and past 30-day use of only cigarettes, only e-cigarettes, only noncigarette combustible products , only smokeless tobacco , and polytobacco .