Aim

To investigate if body mass index causally influences smoking behaviour using a Mendelian randomization approach.

AIMS:

Aim 1: Genome-wide bivariate association analysis of birth weight and key quantitative phenotypes from the 20 year follow-up studies within Raine and ALSPAC.

To date, two genome-wide association studies (GWAS) have been conducted for birth weight (involving the investigators of this proposal), which identified seven single nucleotide polymorphisms (SNPs) that effect birth weight (14, 15). Two of these SNPs are in genes previously associated with type 2 diabetes, ADCY5 and CDKAL1, highlighting the genetic links between fetal growth and postnatal metabolism. At both of these loci, the birth weight lowering allele was associated with increased risk of type 2 diabetes. In addition, the authors investigated a further 47 type 2 diabetes associated variants and found the type 2 diabetes risk alleles at GCK and MTNR2B were associated with higher birth weight. These results indicate that there may be multiple genetic pathways which link birth weight to adult disease. A further link with the DOHAD hypothesis was through the ADRB1 gene, a locus which was associated with birth weight and adult blood pressure. Of the final four SNPs, two were in height associated genes, indicating a continued link with growth over the postnatal period. Although these SNPs found to date for effecting birth weight have pleiotropic effects with risk of disease in adults, the percentage of variance explained by these confirmed loci (0.76%) is lower than many other complex phenotypes (for example, 1.45% for BMI (16), 5.8% for BMD at the femoral neck (17)). Given twin and family studies estimated heritability of birth weight from the fetal genome to be between 10 and 30% (18, 19), there is still a substantial genetic component to be described.

Bivariate genetic association analysis has been shown to have increased statistical power over univariate analysis to detect genetic variants that contribute pleiotropically to the two phenotypes in the opposite direction to the prevailing phenotypic correlation (20-22). This approach is ideal for detecting variants that underlie DOHAD. We will conduct a bivariate genome-wide analysis of birth weight and each of our key phenotypes, with the hypothesis of increasing the statistical power to identify novel genetic variants that have pleiotropic effects.

Aim 2: Gene-based bivariate analysis of birth weight and key quantitative phenotypes from the 20 year follow-up studies within Raine and ALSPAC.

There has recently been a lot of debate regarding the 'missing heritability', or the amount of heritability of common traits that is not accounted for by the known genetic variants (23). As with many common traits, this is seen with birth weight (as outlined above) and each of our key phenotypes. Statistical methods have been developed to estimate the proportion of variability explained by common genetic variants on genome-wide SNP chips (24). By partitioning the genome into smaller chunks of SNPs, these methods can be applied to estimate and test the heritability of a particular genomic region (25, 26). This novel statistical approach efficiently and effectively combines genetic variants across a region to assess whether regions of the genome influence trait values. Utilizing this novel approach and the bivariate structure of the data, we will conduct a bivariate analysis of birth weight and the key quantitative phenotypes by partitioning the genome-wide chip data into groups of SNPs that belong to genes.

AIMS

The aim is to compare online only completion with a choice of online or paper completion in terms of response rates and costs for the YP questionnaire.

Objective 1: To generate and supplement existing genetic, epigenomic and metabolomic data across five large European cohort studies to provide an unparalleled resource for investigating associations between early life factors and subsequent obesity, neurodevelopmental and mental health outcomes.

Objective 2: To integrate inter-generational phenotypic, genetic, epigenomic and metabolomic data across the five cohort studies through harmonization of data and meta-analysis.

Objective 3: To identify causal relationships between prenatal exposures and metabolic and neurodevelopmental outcomes using a variety of state-of-the-art causal analytic methods including Mendelian randomization, maternal-paternal, between-sibling and cross-cohort comparisons.

Objective 4: To synthesise information to inform and prioritise future interventions and public health policy that will have a meaningful and lasting impact upon the incidence of these health challenges.

The primary intent of ROHgen is to investigate the influence of ROH on complex traits; to show for which traits and end points there is evidence of recessive effects, how strong they are and eventually potentially to count and map these effects. This will be done by conducting the same analyses across many cohorts and then sharing various statistics relating to ROH or trait associations for central meta-analysis.

Hypothesis: ROHs are associated with health-related outcomes and quantitative traits of interest.

AIMS:

The aim of this study is to examine whether children with higher autistic trait scores are more likely to experience reduced breastfeeding exposure, and the direction of this possible relationship. We will look at two opposing hypotheses; that reduced breastfeeding duration may be an aetiological risk factor for the subsequent development of autism spectrum disorders, or that an early manifestation of autistic traits is a decrease in the duration of breastfeeding relative to the mother's intended duration.

AIMS

1. To assess whether change in dietary habits from adolescence to young adulthood can modify the gut flora, increase the production of noxious metabolites (TMAO) and accelerate arterial disease progression.

2. To assess whether dietary supplementation of probiotics can modify the gut flora and can benefit the lipid profile and vascular risk of overweight and obese individuals

The aim of the study is to explore the association of distinctive facial features with candidate genes reported to be associated with facial development.

We intend to replicate the findings reported by Claes et al 2014 where the authors found an association of 20 genes with distinctive facial features. This study uses a novel technique bootstrapped response-based imputation modelling to explore the relationships of sex, genomic ancestry and a set of craniofacial candidate genes. The procedure will be replicated on 4747 3D 15 year old facial shells. In addition an additional 30 genes which did not show any association in Claes' sample of 592 will be explored using the ALSPAC cohort.

Essentially the proposed method uses dense spatially dense quasi-landmarks on 3D images (greater than 7,000), principal component analyses and a new partial least squares regression (bootstrapped response-based imputation modelling (BRIM) to measure and model facial variation.

By simultaneously modeling facial shape variation as a function of sex and genomic ancestry along with genetic markers in craniofacial candidate genes, the effects of sex and ancestry can be removed from the model thereby providing the ability to extract the effects of individual genes.

Although the combined impact of genetic variants on phenotypic traits has become of increased interest over the last few years, there has so far been little research undertaken to understand the combined impact of variants on DNA methylation. We intend to apply collapsing approaches to collapsed methylation scores to investigate this. This could be an attractive approach in contrast to individual CpG site analyses for many reasons, such as computional efficiency, reducing the burden of multiple testing, smoothing out artefacts and representing the underlying biology in a different form by collapsing across functional units (e.g. CpG islands, including/excluding shores, entire promoter regions). Furthermore, the manner in which methylation data is collapsed is an aspect of this work which will need to be evaluated. There are currently some methods out there which could be used for this purpose, although so far they have mainly been used to investigate the relationship between methylation and phenotypic traits.

As pairwise association analyses for individual CpG sites is currently being undertaken using the 450K methylation data in ALSPAC, this should provide a useful comparison for this proposed approach.