Significance testing for small annotations in stratified LD-Score regression

• DOI: 10.1101/2021.03.13.21249938
• Publication Date: 2021-03-24T18:00:15Z
• Authors: Katherine C. Tashman; Ran Cui; Luke J. O’Connor; Benjamin M. Neale; Hilary K. Finucane
• Keywords: Jackknife resampling

Abstract S-LDSC is a widely used heritability enrichment method that has helped gain biological insights into numerous complex traits. It primarily been to analyze large annotations contain approximately 0.5% of SNPs or more. Here, we show in simulation that, when applied small annotations, the block jackknife-based significance testing does not always control type 1 error. We inflation error for due both noisiness jackknife estimate standard and non-normality regression coefficient estimates. use percent 0.01 centimorgan blocks genome overlapped by annotation quantify size an extent which cluster together, find thresholds on this value above controlled. have implemented test LDSC software informs users they compute LD scores if pass threshold producing controlled Author Summary Genetics rapidly evolving field allows us link our genetic code physiological manifestations disease. A key part work finding regions contribute disproportionately underpinnings commonly tool provide such insight stratified score (S-LDSC). how much set genomic contributes overall phenotype, whether more than would expect chance. Here apply regions, it give accurate chance phenotype. characterize what means be “small” restrict prevent false positive results.This helps ensure as continue pursue analyses at scale, report only truly significant results will help further understand etiology many traits study.