Multi-QMDR

Detect gene-gene interaction/epistasis by using multivariate phenotypes and MDR

Notice

All source codes were listed in file "Multi-QMDR.R" for implemeting Multi-QMDR

Citation

Yu, Wenbao, M-S. Kwon, and T. Park. "Multivariate Quantitative Multifactor Dimensionality Reduction for Detecting Gene-Gene Interactions." Human heredity 79.3-4 (2015): 168-181.

Example

Try run_example.R to get a quick start, and an example data was provided too

Usage

In R:

source("Multi-QMDR.R")

multi_qmdr(phes, snp.mat, method = 'FPC', K = 2, nperm = 0, test.type = 'ht2', 
           sele.type = 'cvc', kfolds = 10, covrt = NULL)

• inputs:

  • phes ---- phenotypes, n times d matrix
  • snp.mat ---- snp matrix, n times p matrix
  • method ---- one of 'FPC', 'WPC', 'WSPC'; if you just use single phenotype (qmdr), specify method=i, which will use qmdr and the ith column of phes as phenotype; defalt 'FPC'
  • K ---- K-way interactions, default 2
  • nperm ---- permutation times for calculating pvalue for the best model (0 if pvalue if not needed; default)
  • test.type ---- test statistics, could be 'ht2' or 't', corresponding to hotelling t2 test and t test; default 'ht2'
  • sele.type ---- the way to tune the best model, 'cvc' or 'score'; default 'cvc'
  • kfolds ---- k-fold cross validation; default 10
  • covrt ---- the covariate matrix; default NULL (no covariates)

• output: a list with elements as follows

  • best_ksnps ---- the snp ids for the best model
  • cvc ---- the cvc number of the best model (out of 10)
  • score ---- the test statstics for the best model
  • pvalue ---- the corresponding empirical pvalue for the best model