The primary aim of this study was to investigate the association between genetic variants and childhood growth trajectories across the Early Growth Genetics (EGG) consortium. BMI trajectories in childhood tend to be difficult to model due to the complexities of growth. To conduct analysis on a genome-wide scale, the analysis of each SNP must be computationally efficient. Additional methods for analysing these data are currently being investigated including the Super Imposition by Translation and Rotation (SITAR) method, an extension to the LME based on a multivariate t distribution to account for the increasing heteroscedasity and a semi-parametric mixed model. Variables of interest: * BMI at all available time points between 1 and 16 years of age * Age and Sex for adjustment and stratification in analytic models * Singleton/Multiple birth status and ethnicity for exclusion of related individuals and non-Caucasians from analysis * 6 previously published adult BMI associated polymorphisms to assess the ability to detect genetic associations using the different modelling frameworks.

Analyses: (Inclusions/Exclusions: Include only singletons, Caucasian ethnicity only, BMI measured at 2 or more time points between 1 and 16 years.) Stage One: Analysis steps: 1. Run analysis of chosen longitudinal methods in all three cohorts 2. Include each BMI associated SNPs individually to the models in all three cohorts. 3. Compare methods focusing particularly on model fit, computational time, generalizability and ability to detect genetic associations. Choose most appropriate method for further analysis. 4. send chosen analyst from each cohort the code to conduct GWAS analysis using longitudinal method or the summary statistics derived from longitudinal method for analysis as a quantitative trait against the GWAS. 5. Compare the top genes from the meta-analysis of the GWAS results from the three cohorts to those detected in the "GWAS on infant and early childhood growth parameters until age 7 years" study to ensure we are investigating two separate phenotypes. Stage Two: Once the feasibility of pursuing this proposal and the proposal named above as independent investigations has been determined using these three cohorts, we will extend this analytic plan to other cohorts willing to participate and with relevant data in EGG. Stage Three: The most strongly statistically significant genetic loci will be followed-up with more intensive analyses using other phenotypes indicating obesity (e.g. waist circumference and waist-to-hip ratio) and other methods developed during previous work modelling BMI trajectories in EGG cohorts