Multicollinearity in graphical lasso

[Daniel Gingerich]

As I learn more statistics I am beginning (I think) to understand the graphical lasso.   One thought came to mind. 

Does multicollinearity in the ATAC peaks affect the graphical lasso?  In terms of measuring the partial correlation of peak1 and peak2, given all other peaks, if a set of background peaks are correlated, this would not affect the estimate of X or Y.  Thus the partial correlation is unaffected.  

I kind of answered my own question, but want to see what your thoughts are, as I am still trying to understand. 

[hpl...@gmail.com]

I'm not sure I entirely understand you question, but yes, the idea with graphical lasso (lasso in general) is that it will choose a reduced set of covariates which hopefully gets you a better estimate of underlying correlation. In the cicero context it should help reduce the effect of correlated background peaks.