METACARPA performs scalable meta-analysis between genetic association studies, both effect-size based and p-value based, while correcting for unknown sample overlap.
Meta-analysis allows to summarize association results across several studies, but assumes independence of samples. Lin and Sullivan (2009) have developed a method that accounts for overlapping samples when the degree of overlap is known. Province and Borecki (2013) use the tetrachoric correlation of transformed p-values to estimate relatedness in a p-value based meta-analysis. We implement this method and adapt this estimator to Lin and Sullivan’s to provide a correlated meta-analysis of both p-values and effect sizes.