D2Ks: Convert pairwise distance matrix to kernels

In baolinwu/MiRV: Kernel based independence test for microbiome association studies

Description

Adapted from the MiRKAT package (Zhao et al., 2015). Modified for better efficiency avoiding unnecessary matrix products, and output kernel and its eigen decomposition for followup analysis. Can show K_{ij}= -0.5(D_{ij}^2-\bar{D^2}_{i,}-\bar{D^2}_{.,j}+\bar{D^2}), where \bar{D^2}_{i,},\bar{D^2}_{,j},\bar{D^2} are row, column and global averages. For symmetric distance matrix, \bar{D^2}_{i,}=\bar{D^2}_{,i}.

Usage

D2Ks(D)

Arguments

D	pairwise distance matrix

Value

K

computed (non-negative) definite kernel matrix

Kh

computed from eigen decomposition of K, such that K=K_hK_h^T

Km

the effective rank of K

References

Chen, J., Chen, W., Zhao, N., Wu, M.C., Schaid, D.J., 2016. Small Sample Kernel Association Tests for Human Genetic and Microbiome Association Studies. Genet. Epidemiol. 40, 5–19.

Zhao,N. et al. (2015) Testing in Microbiome-Profiling Studies with MiRKAT, the Microbiome Regression-Based Kernel Association Test. AJHG, 96(5): 797<e2><80><93>807.

Guo,B. and Wu,B. (2017) On the fast small-sample kernel independence test for microbiome community-level association analysis. Biometrics, doi:10.1111/biom.12823.

baolinwu/MiRV documentation built on May 28, 2019, 6:37 p.m.