1. Background

The aim of this study is to perform a meta-analysis of GWAS results on gender role behaviour in young children as assessed by male, female and overall gender Pre-School Activities Inventories (PSAI) scores1. Although gender role behaviour increases during childhood, especially during preschool years, the ranking of sex typed behaviour has been shown to be consistent when assessed at 2.5, 3.5, 5 and 8 years2. Gender role behaviour is heritable (Pre-school children: h2 Boys = 0.34; h2 Girls = 0.573) and may relate to underlying quantitative trait loci (QTL) that are accessible through genome-wise association studies (GWAS). This proposal aims to meta-analyse gender role behavior in children at age 3 and 5 using male, female and overall gender PSAI scores1 that are available in ALSPAC and the Twins Early Development Study.

2. Traits of interest

• Male gender play score (MGPS)
• Female gender play score (FGPS)
• Gender PSAI score (PSAI)

3. Participating studies (as of 27-03-2010): Individuals with GWAS and phenotype data

* ALSPAC (MGPS/FGPS/PSAI 42m: 3022; MGPS/FGPS/PSAI 42m 57m:3023/3012/2980)

* TEDS (MGPS/FGPS/PSAI 3y: ~4000; MGPS/FGPS/PSAI 4y: ~4000)

4. Total projected number of subjects (as of 27-03-2010)

* MGPS/FGPS/PSAI (36m to 42m) ~ 7000 individuals

* MGPS/FGPS/PSAI (48m to 57m) ~ 7000 individuals*

5. Genotyping + Imputation

* Genotyped SNPs (Illumina 317k, 610k)

• Sample-QC: Individuals removed based on missingness, heterozygosity, relatedness, population and ethnic outliers, and other cohort-specific QC steps.

* For stratification analysis, please generate principal component scores (or MDS) using for example EIGENSTRAT. These will be included as covariates (first 5 PCAs).

* Imputation based on HapMap Phase II CEU SNPs.Preferred release 22 of HapMap, build 36.

• SNP-QC before imputation: SNPs removed based on missingness (SNP-call greater than 95%), minor allele frequency (MAF less than 1%), Hardy-Weinberg (HWE p-value less than 5E-07);
• o QC in terms of imputation quality is needed for imputed SNPs.

6. Association analysis

* Assuming unrelated individuals

* Assuming White European samples

* Assuming autosomes only

* Assuming all SNPs are aligned to Hapmap Phase II + strand

* Perform a linear regression for

• MGPS [36m;42m]
• FGPS [36m;42m]
• PSAI [36m;42m]
• MGPS [48;57m]
• FGPS [48;57m]
• PSAI [48;57m]

Complete sample

a) Trait = SNP + Age + Age2 + first 5 PCA

Sex-specific analysis (Males)

b) Trait = SNP + Age + Age2 + first 5 PCA

Sex-specific analysis (Females)

c) Trait = SNP + Age + Age2 + first 5 PCA

We may consider rank-transformation for bimodal PSAI scores.

* Perform association analysis on the dosage score (Additive model): SNP coded as allele dose from 0 to 2 on imputed data only

* Use analysis software that accounts for genotype imputation uncertainty (as included in MACHQTL, ProbABEL, SNPTEST), adjusting for population structure and covariates.

* Analyse all imputed SNPs,no filtering on call rate/HWE/MAF/imputation quality (performed on meta-analysis stage)