README.md

In baolinwu/MetaSAT: Meta-analysis of variant-set association tests

MetaSAT

• An R package for meta-analysis of variant-set association test

Meta-analysis models

• We study meta-analysis methods that can account for varying levels of heterogeneity of genetic effects both within and across studies.
• Burden type test (BT): similar variant effects
• Variance component test (VT): heterogeneous variant effects
• Adaptive test (AT): adaptively combine BT and VT
• Fixed-effects (FE) model
• Assumes homogeneous effects across studies.
• Heterogeneous-effects (HE) model
• heterogeneous effects across studies and across variants within each study
• Robust heterogeneous-effects (RHE) model
• heterogeneous effects across studies but homogeneous effects across variants within each study
• In summary, a total of four general tests
• FE BT: homogeneous effects across and within studies.
• FE VT: homogeneous effects across studies/heterogeneous effects within studies
• HE VT: heterogeneous effects across and within studies.
• RHE BT: heterogeneous effects across studies/homogeneous effects within studies
• Four ATs are implemented
• FE AT (FAT): combine FE BT and FE VT
• HE AT (HAT): combine FE BT and HE VT
• RHE AT (RAT): combine RHE BT and HE VT

• BT based AT (BAT): combine FE BT and RHE BT

• Ref

• Wu,B. and Zhao,H. (2018) Efficient and powerful meta-analysis of variant-set association tests using MetaSAT. tech rep.

Sample R codes

• The MetaSAT R package also contains the summary statistics and their associated asymptotic covariance matrices for 9 variants across 3 studies as discussed in Section 4 of the tech document. They can be used to reproduce the analysis results in the manuscript.

 data(WZda)
 Us = WZda$Us; Vs = WZda$Vs
 FESAT(Us,Vs)
 HESAT(Us,Vs)
 RESAT(Us,Vs)
 RBAT(Us,Vs)

baolinwu/MetaSAT documentation built on May 13, 2019, 5:22 p.m.