Generalized association study for matched non-Gaussian data The GAS_matlab folder contains main MATLAB functions for implementing the proposed methods under the Generalized Association Study (GAS) framework. It provides a tool box for flexibly analyzing the association between two data sets with heterogeneous data types.
Sim_Setting.m The m file contains all the simulation settings used in the paper.
Sim_RankEst.m The m file provides a demo showing how to use the main functions to estimate the latent ranks of heterogeneous matrices.
Sim_ParamEst_dense.m The m file provides a demo showing how to use the proposed methods to estimate model parameters.
Main functions in GAS_matlab:
GAS.m The function estimates the joint and individual structure in the GAS model. It contains several variants such as the one-step approximation, and the sparse estimation.
GAS_PermTest.m The function conducts permutation tests to evaluate the statistical significance of the association coefficient.
ExpPCA.m The function implements the exponential family PCA method.
Nfold_CV_Mixed.m The function conducts N-fold cross validation to select the best latent rank for dual-typed data.
Nfold_CV_Single.m The function conducts N-fold cross validation to select the best latent rank for a single data matrix.
Contact: Gen Li, PhD
Assistant Professor of Biostatistics, Columbia University
Email: gl2521@columbia.edu
CopyRight all reserved
Last updated: 2/11/2017