This protocol uses a window of 1500 variants, shifted by 10% for each new round of comparisons, and a threshold of R 2  > 0.2. The window size of 1500 variants corresponds to the large, high LD chromosome 8 inversion, while the shift of 10% represents a trade-off between efficiency and thoroughness

It is necessary to remove rare variants from GWAS because the certainty of the genotype call is reduced by their low minor allele count. Even in common variants, however, genotyping and genotype recalling are subject to technical error, with the result that a proportion of variants and samples are of low quality, and should be removed from the analysis.

For the smallest studies, where fewer than 1000 individuals are investigated, a cut-off of 5% should be considered—this is in line with the analysis program GenAbel, for example, which uses a minor allele count of 5 as its cut-off [ 18 ].

The expectation is that IBD = 1 for duplicates or monozygotic twins, IBD = 0.5 for first-degree relatives, IBD = 0.25 for second-degree relatives and IBD = 0.125 for third-degree relatives. Due to genotyping error, LD and population structure there is often some variation around these theoretical values and it is typical to remove one individual from each pair with an IBD > 0.1875, which is halfway between the expected IBD for third- and second-degree relatives. For these same reasons an IBD > 0.98 identifies duplicates.

The method works best when only independent SNPs are included in the analysis. To achieve this, regions of extended linkage disequilibrium (LD) (such as the HLA) are entirely removed from the dataset8 and remaining regions are typically pruned so that no pair of SNPs within a given window (say, 50kb) is correlated (typically taken as r2>0.2)

Our primary aim was to generate a cohort large enough to examine the heritability of prognostic therapy outcomes. However, the meta-analysis estimate of SNP heritability was low and non-significant (h2SNP = 0.09, SE = 0.17). A sample size of 2724 has 80% and 99% power to detect a SNP-heritability of 33% and 50%, respectively94. To achieve 80% power to detect a heritability of 20%, a sample of 4500 individuals will be required. A meta-analysis of 2 799 individuals was sufficient to detect a significant heritability estimate for therapy outcome to antidepressant drugs (h2SNP = 0.42, SE = 0.18) and this was the first evidence of a genetic component for treatments outcome of any kind

The meta-analysis sample (n = 2724) had 80% power to detect variants explaining 1.5% of the variance and 42% power to detect variants explaining 1% of the variance. Therefore, it is not especially surprising that we do not detect any variants at genome-wide significance. Typically, GWAS of psychological traits have required tens of thousands of participants to detect SNPs at genome-wide significance

which is the basis of our planned second release (PLINK 2.0).

A population genetic interpretation of GWAS findings for human quantitative traits

Vulnerabilities of transcriptome-wide association studies

Transcriptome-wide association study of schizophrenia and chromatin activity yields mechanistic disease insights

Introduction to Population Genetics

Genome-wide methylation data mirror ancestry information

Analyzing Copy Number Variation using SNP Array Data: Protocols for Calling CNV and Association Tests

CNVassoc: Association analysis of CNV data using R

Rare copy number variant (CNV) data

A New Method for Detecting Associations with Rare Copy-Number Variants

Additional information: