Department of Biostatistics

SVSI: Fast and Powerful Set-Valued System Identification Approach to Identifying Rare Variants in Sequencing Studies for Binary and Ordered Categorical Traits

We have developed the SVSI package that attempts to test genetic associations with complex phenotypes such as ordered categorical phenotype or binary phenotype based on set-valued model and using a novel Set-Valued System Identification algorithm. The set-valued model models the classification/dichotomization process of a continuous complicated unobserved or unobervalble phenotype and models the relationship between categorical outcome and genetic variants. Set-Valued System Identification (SVSI) algorithm is a new algorithm based on expectation-maximization (EM) which has been proposed and appears in a wide range of applications such as sensor networks and telecommunications. The method is being published by Kang G et al (2014) in Annals of Human Genetics.

The method is implemented in SVSI R package written for the freely available statistical computing language R.

To Install SVSI:

  • Download SVSI 0.1.tar.gz and execute the command install.packages ("download.dir/SVSI_0.1.tar.gz",repos=NULL,type="source") in R, where download.dir is the directory to which the file was saved.
  • The SVSI user’s manual explains how to use the SVSI package.
  • To receive alerts regarding software updates, send an e-mail with "Updates on SVSI Software" in the subject line.

More questions? Contact Guolian Kang.