How to enter covariates/confounding variables for CWAS analysis

Submitted by jwhnavy on

Dear Dr. Yan,

1) Could you explain how to enter covariates/confounding variables when performing CWAS analysis using DPARSFA? I'd like to generate the CWAS map for specific behavioral performance (a covariate of intrest) while controlling for potential confounding variables (e.g., age and gender). I made a text file consisting of 3 columns: a covariate of interest (for lst column), age (for 2nd column), and gender (for 3rd colume). Is the confounding effect of age and gender on the generated CWAS map controlled if I use the text file mentioned above? 

2) While Shehzad et al., (2014) used "sqrt(2(1-r))" as the distance measure, I found "(1-r)" as the distance measure in y_CWAS.m file. Is "1-r" used to calcuate the distance when using CWAS on DPARSFA? 

Best regards, 

Wi Hoon 

YAN Chao-Gan

Thu, 08/18/2016 - 00:57

Hi Wi,

The CWAS in DPARSF hasn't integrated the function of covariates yet. 1-r is used.

Best,

Chao-Gan

jwhnavy

Fri, 08/19/2016 - 01:37

In reply to by YAN Chao-Gan

Thank you for your reply. 
 
I am wondering if there is any way to control confounding effects (i.e., age and gender) when using CWAS implemented in DPARSFA. 
 
1) How about using residual brain images and residual regressors after regressing out confounding variables (i.e., age and gender) at voxel level across subjects as inputs for CWAS implemented in DPARSFA ?
 
2) Is there any way to control confounding effects by simply adding some code in “y_CWAS.m” file? I wonder how to enter confounding variables (e.g., age) for each subject in MDMR because each element in a distance matrix is based on a pair of connectivity between two subjects (maybe average age across two subjects?). 
 
I look forward to your advice. 
 
Best, 
Wi Hoon